Full Article Archive
Probe Back Issues
CTKAformerly published Probe Magazine. Most of the articles on this site first appeared in Probe.
If you would like to submit an article to be considered for publication on this site, please send mail to us at here.
Jim DiEugenio's Upcoming appearances and radio Interviews:
April 13th, Barnes and Noble, Metro Pointe,
901 B South Coast Drive Ste 150, Costa Mesa,
May 4th, Barnes
and Noble, Orange Town & Country
791 South Main Street Suite 100,
NEW DATE! May 18th, Barnes
and Noble Bookstore in Manhattan Gateway Shopping Center 1800 Rosecrans
Avenue Building B, Manhattan Beach, CA 90266
October 16-19th Passing the Torch Conference, at Duquesne University in Pittsburgh
November 21-24, November in Dallas, at the Adolphus Hotel in Dallas
JFK: The French Connection, by Peter Kross Review by Seamus Coogan
on Lunch with Arlen Specter on January 4, 2012
KENNEDY & ME: A Very Good Book With A Few Pages of Trouble
Jim DiEugenio analyzes and summarizes Larry Hancock's
interesting and unique new book Nexus:
The CIA and Political Assassination
Jim DiEugenio reviews the work of Chris Matthews on the life and death of President Kennedy, including his latest biography, "Jack Kennedy: Elusive hero".
IN DALLAS: LBJ, the Pearl Street Mafia, and the Murder of President
The Connally Bullet Powerful evidence that Connally was hit by a bullet from a different assassin, by Robert Harris
Joseph Green on the late Manning Marable's new full scale biography of Malcolm X.
JFK and the Majestic Papers: The History of a Hoax by Seamus Coogan
- and -
Wikipedia? by JP Mroz and Jim DiEugenio (3 part series)
is Anton Batey?
Exclusive excerpts from Mitchell Warriner's long
awaited new book on
Hear No Evil: Social Constructivism and the
Forensic Evidence in the Kennedy Assassination
|HSCA||Thomas||Source of Shot|
According to Thomas, the “noise” remains unidentified, although JBC apparently responded to it. It did not have the usual characteristics of a gunshot (i.e., no typical echoes). The rogue shot, according to Thomas, came from the rear, but its exact origin, and even its destiny, are not known—which is where the third gunman might sneak into Thomas’s scenario. Thomas cites Z-224 as his key shot: it establishes the critical interval (along with Z-313) that ties together the audio and visual evidence for him, and it also defines the moment of the SBT. Unfortunately for Thomas, even the HSCA does not recognize this moment (Z-224) as a shot. In other words, the HSCA does not support two of Thomas’s chief claims for Z-224: the moment of the SBT and first part of the critical time interval.
Myers radically disagrees with Thomas about the Z-frame that matches the Hughes frame (near the end of the Hughes film)—that’s the one that shows McLain rounding the corner at Main and Houston. Myers claims that it was taken at about Z-135, just 2.19 seconds before the first suspect shot. Thomas, on the other hand, claims it was 3.8 to 5.9 seconds before the first suspect shot. In Thomas’s scenario, McLain needs lots of time to reach the site of the first suspect shot, which was 180 feet away, and which would have required a speed of over 20 mph during this interval. (Thomas does address these issues in some detail on pp. 673-684).) As further support for his case, Myers adds that the film record shows that Camera Cars 1, 2, and 3 all stopped in front of the TSBD at about six seconds after the final head shot— and they remained stationary for about 15 seconds.
My Summary. McLain was not located where the HSCA (or Thomas) needed him. The suspect shots on the Dictabelt do not match to McLain. Oddly enough, the HSCA seriously disagreed with Thomas on multiple points.
Michael O’Dell’s presentation (see my Appendix 4 for a citation) is a model of clarity. Beginners might well start with his article. He also provides a link to the National Academy of Sciences website for listening to these digital CDs:
(However, I could not access this site.) He reminds us that both police recording devices (for Ch-1 and Ch-2) engraved a track on a plastic medium. The Dictabelt (Ch-1) used a rotating cylinder while the Audograph (Ch-2) used a flat disc,like a phonograph record. O’Dell emphasizes that Thomas’s case is based on a continuous recording (during the critical interval) on Ch-2, i.e., Thomas can permit no missing time on Ch-2 for his scenario. To support his fundamental assumption, Thomas had relied on the initial NRC data (based on the Bowles copy). But O’Dell found a problem with this: he notes that the needle assembly on the Audograph does not move. Instead the disc (mounted on a spindle) moves horizontally under the fixed needle. Because this needle is somewhat flexible, though, repeats (overdubs) can occur for one or even two rotations. But the main point is this: because the horizontal movement (of the disc) never stops, the fixed needle must eventually catch up to where it should be—i.e., at the end, the time lost (or gained) is either nil or very small. The NRC, however, had not taken into account the forward skips. Therefore, when the NRC corrected for the overdubs, they got the wrong answer. In fact, it would have been better if they had not corrected at all—since the correction was nearly zero (or actually zero). O’Dell claims that the FBI copy of Ch-2 (made in 1981) is much better—it has no skips or repeats (overdubs). Therefore, the Bowles copy (used in the original NRC report, but not used in their follow-up report) should not be relied upon. The Bowles copy, unfortunately, is the one that Thomas had used. Instead, the FBI copy should be used for timing issues. (That’s the one that Linsker, et al. used in their follow-up report.) There is only one problem with this though: this recording could not be played back on the Gray Audograph. The reason is that the device kept producing skips and repeats on playback. (Any recording, for accuracy, ideally should be played back on the same machine that recorded it.) Therefore, a standard phonograph table was used instead, but this created new problems. Phonograph tables operate at a constant rpm, but the Audograph did not. Instead the Audograph rpm decreased as the recording progressed. As a consequence, when the Audograph disc is played back on a standard phonograph table, the sounds at the beginning are slowed down (time intervals are longer and frequencies are lower), whereas the later ones are accelerated. So a method had to be devised for correcting this. The NRC used the 60 Hz hum in the background, which had been introduced during the original recording. O’Dell displays a remarkably linear graph of frequency vs. time, which shows that this correction can be done very accurately. Ch-1, on the other hand, does not have this problem, nor does it have skips or repeats; it is only important to get the right playback speed for Ch-1 (which can be done). So, the bottom line is that the original NRC timeline was wrong, and needed to be corrected to obtain the right synchronization between channels.
O’Dell displays (in his Table 1, copied just below) the corrected time intervals. Note especially these intervals (according to O’Dell’s table) for Ch-2:
|HOLD to CHECK (labeled ALL by Thomas)||130.68 sec|
|CHECK to YOU||12.23 sec|
|YOU to ATTENTION||90.84 sec|
O’Dell and Thomas, unfortunately, sometimes use different labels for the same event, which can be very confusing. For example, “I’ll CHECK ALL these motorcycle radios” is called CHECK by O’Dell, but is labeled as ALL by Thomas. Adding to the confusion, Thomas labels Fisher’s supposed phrase (just seconds before the suspect shots—“I’ll check it”) as CHECK. In this discussion, since it is based on O’Dell’s work, I shall stick with his labels. We next compare the above intervals to those listed by Thomas (in his Table 14, p. 640). However, another problem emerges here: Thomas appears to have reversed the two very close utterances by Bellah: YOU and ALL (the latter is labeled CHECK by O’Dell). This mistaken reversal is confirmed not only by O’Dell, but also by the Bowles transcripts (in Savage’s book, p. 406). After correcting for this, we can present Thomas’s intervals for Ch-2 (using O’Dell’s notation for clarity):
|HOLD to CHECK||
|CHECK to YOU||
|YOU to ATTENTION||
The reader will immediately see that O’Dell (130.68 sec) and Thomas (143 sec) disagree about the 1st interval, whereas they nearly agree on the 2nd and 3rd. Also notice that Thomas claims (in the legend for his Table 14) to have used “corrected” playback intervals; he also cites “Data from O’Dell (2003).” (I could not find a date on O’Dell’s essay, but Thomas provides strong evidence that he and I have cited the same essay by O’Dell.) So they should agree on this 1st interval. Neither O’Dell nor Thomas have addressed this apparent disagreement (of 143 – 130.68 = 12.32 seconds).
The NRC panel had shown that Decker’s phrase, “...HOLD everything secure...” appeared as crosstalk on Ch-1 at the same time as the suspect 3rd and 4th shots. (The 1st suspect shot occurred 6.5 seconds before that.) O’Dell next discusses how that conclusion was reached. His Table 2 (copied here) provides the supporting data for that conclusion.
(Times are relative to "HOLD")
Ch 2 Event
Ch 1 Event
"GOING to the..."
"CHECK all these…"
"CHECK all these…"
"YOU want me…"
"YOU want me…"
Notice that “GOING to the hospital” occurred 67.5 seconds before HOLD (on Ch-2). The 1st suspect shot occurred 6.5 seconds before HOLD (on Ch-1). If HOLD is a valid crosstalk, we can then conclude that “GOING” occurred 61.5 seconds (67.5 – 6.5 = 61.5) at least before the 1st suspect shot. That makes no sense, since the “GOING” utterance presumably describes action taken after the shooting (probably after all the actual shots had been fired). The case is closed therefore—the suspect shots cannot represent the actual shooting. O’Dell provides a useful timeline (copied here) for the four cases of crosstalk, including the final deliberate simulcast, in his Figure 2. (Thomas had added a fifth candidate, Fisher’s CHECK, in his book, so that one does not appear in this figure.)
How much time, if any, was missing on Ch-2? Thomas argues for almost none. If he is wrong about this, though, then his argument for synchrony is also wrong, and his entire case fails. This is a crucial point. His argument is based on a graph (see his Figure 16.3) of dispatcher’s announced time vs. playback time. Thomas argued that this was a straight line and that the slope was nearly 1.0. Even if this were true, though, the margin of error (±0.05) is large enough that some missing time might exist. Therefore, Thomas’ case is already weaker. However, Thomas’ problem is worse than that: according to O’Dell, the slope is not 1.0, but actually closer to 0.918 (0.94 according to Linsker). Therefore time is almost certainly missing on Ch-2—and that seriously undermines Thomas’s basic assumption. In fact, a slope of 0.918 would correspond to about 30 missing seconds during that total time interval (of about six minutes)—and that is significant. As further support for this conclusion, O’Dell emphasizes that Ch-1 has about 28 more seconds than Ch-2 between two specific crosstalk samples: “...Hold everything secure...” and “I’ll check all these motorcycle radios.” That implies to him that Ch-2 stopped for 28 seconds. Between “I’ll check all these motorcycle radios” and “You want me” we see another difference (of 2.4 seconds); between “You want me” and “Attention” is another difference (24.8 seconds). (From Table 14 by Thomas, p. 640, we see similar differences: 31, 3, and 24 seconds, respectively. In each case, less time is apparent on Ch-2 (than on Ch-1), which is consistent with the conclusion that Ch-2 had stopped.) So, if one accepts Thomas’s thesis that Ch-2 did not stop and the crosstalk misalignments were caused by needle skipping (or something else with a similar effect), then just one skip could not explain this. Instead, at least one skip must occur between each example of crosstalk (because these differences appear for each interval shown above). Such persistent and regular skipping seems unlikely; on the other hand (Thomas won’t like this), it’s easier to conclude that Ch-2 did stop, as it was supposed to do (after four seconds of silence).
Then Thomas proposes another option: synchronizing on “You want me,” instead of “...Hold everything secure...”. He thought that these two cases were inconsistent and that the former would more reasonably place the suspect shots before Curry’s “Go to the hospital” (GO). Thomas said this scenario could be explained in one of two ways: (1) “...Hold everything secure...” was misplaced on Ch-1 or (2) Ch-2 stopped between these two utterances. But here is O’Dell’s response (based Thomas’s scenario): on Ch-1, there are 181 seconds between the first suspect shot and “You want me,” while on Ch-2 there are 210 seconds between “Go to the hospital” and “You want me.” Therefore, even Thomas’s new synchronization places GO at 29 seconds (210 - 181 = 29) before the first suspect shot. Linsker (see below) makes a similar case.
Finally, O’Dell makes a clever suggestion. Using an example of speech, he shows that simple enunciation of the word, “secure,” with its prominent K sound, produces an impulse pattern quite similar to a gunshot sound. He also claims that the consonants D, T, S (in the right combinations), likewise can do this. This is relevant because it is likely that the suspect shots are actually superimposed on speech sounds. But there is a counterargument here, which I find very compelling: the initial sound (of each suspect gunshot) is highly compressed, due to the AGC. When these sounds are expanded in time, it is still very difficult to display all of the information in them. But that is not the case for human speech, where the information can usually be displayed. If this argument carries the day, as it probably should (O’Dell does not address this counterargument), then we can conclude this: the suspect gunshot sounds are not solely human speech, but rather something else. (Actually, to be fair to O’Dell, he suggests that human speech is superimposed over the suspect sounds.) Of course, that the sounds are “something else” does not prove that they are gunshots either. But something very loud, likely much louder than human speech, must have produced these sounds.
My summary. O’Dell provides a lucid exposition of events. He clarified the issue of skips and repeats on Ch-2 (which essentially cancelled one another). He concludes that the suspect shots came too late and he suggests that the consonants K, D, T, and S contributed to the suspect sounds.
Bugliosi provides a useful summary of the case, including several items not easily found elsewhere. To begin, he notes that Barger emphasized that the human ear could not identify the suspect shots on the Dictabelt. Bugliosi then discusses the 15 matches and six false alarms. I would emphasize that false matches are indeed possible—six of them in this case—even though they meant nothing. BBN had expected 13 such false alarms, but only six had been found. In view of this, the BBN stated, “It is not unreasonable to expect that there are seven more [false alarms]...”. They concluded therefore that, of the remaining nine correlations, each was “...equally likely to represent a detection or a false alarm” (8HSCA 95,106).
Barger testified that, at the time of the first suspect shot, McLain was about 120 feet behind the presidential limousine; i.e., he was on Houston St., about ten feet from Elm (2 HSCA 68, 101). But no photograph has been found to document this. Based on Barger’s findings, the HSCA said that the first shot was fired at Z-168. WA, however, instead chose Z-151 to Z-161. They disagreed about the second shot, too: Z-197 (BBN) vs. Z-188 to Z-191 (WA). They even disagreed about the GK shot: Z-304 (BBN) vs. Z-295 to Z-296 (WA). (See HSCA Record 180-10110-10234, Draft of the HSCA Report, December 13, 1978, p. 54; HSCA Report, p. 80.) These disagreements are usually greater than five frames; the reader should recall that this small interval (of five frames) was touted by Thomas as an excellent match between the audio and visual data. That the two acoustical teams consistently disagreed by more than five frames makes me wonder if Thomas just got lucky.
Anthony Pellicano, president of Voice Interpretation & Analysis Ltd. in Chicago, concluded that the stuck microphone was not part of the motorcade and therefore, “the noise impulses detected ...were not shots.” His conclusions were largely based on the siren sounds, which only appear about two minutes after the suspect shots. On Ch-2, sirens are audible in the background of Curry’s words, “Go to the hospital.”
The police radio has both a receiver and transmitter. The transmitter has a single button, which must be pushed to talk. The receiver has two possible positions: Ch-1 and Ch-2. Bowles claims that the receiver switch never got stuck; only the push-to-talk button did.
Bowles adds that the HSCA never analyzed the motorcycle sounds, i.e., they never tried to determine whether the stuck microphone was on a two wheeler vs. a three-wheeler. Bowles insisted that the sound difference (between two and three wheels) was dramatic, like comparing a new Lincoln to a Model-T.
Bowles played portions of the test tape (i.e., the HSCA test shots) to Bugliosi. He (presumably meaning Bugliosi) heard voices saying,
“Stand by just a second,” then, “Give us ten seconds when you’re ready” (followed by another voice saying, “Ten-four”), and then, “Ten seconds.” After this there are no voices for ten seconds and then you clearly hear the sound of gunshot, even the echo, on the tape. This sequence was repeated on the tape several times, all followed by a clearly audible shot.
Bugliosi emphasizes that the HSCA used its own microphones and recording equipment, even though they could have used the original DPD equipment; Bowles said the DPD had the same audio systems in 1978 as in 1963. Bugliosi asked Blakey if he could hear the gunshots on the test tapes; his answer was, “Of course we could hear the rifle sounds on our tape.”
Bugliosi names two DPD suspects for the stuck microphone: Willie Price and Leslie Beilharz. He ultimately declines to select one, but does note that Beilharz admitted that he always whistled, even during 1963. Beilharz also recalled being on the Stemmons Freeway, en route to the Trade Mart (consistent with Thomas’s quote above). He even admitted that his microphone may have been stuck for some time, which did not disturb him because he had already received his orders. Most interestingly, Beilharz rode a three-wheeler that day.
WA assumed that Barger’s placement of the stuck microphone in Dealey Plaza was correct, even though Barger admitted that photographic confirmation was missing.
Bugliosi quotes in some detail from the HSCA about where McLain was located with respect to the suspect shots, and concludes that McLain would indeed have had to accelerate in order to make up the required distance (to match the microphone site where the suspect shots occurred). Before McLain started north (toward Elm) on Houston, the Hughes film shows that the presidential limousine had already turned the corner at Houston and Elm. Other films (Turner, Bell, and Zapruder) show the limousine accelerating after completing that turn. McLain appears to confirm this:
Yes, they [the presidential and vice-presidential limousines] were just turning the corner onto Elm St as I came around the corner off Main St (5 HSCA 641).
To close their case, the HSCA stated:
at the time of the assassination he [McLain] would have been in the approximate position of the open microphone near the corner of Houston and Elm indicated by the acoustical analysis (HSCA Report, p. 76).
In fact, McLain had said no such thing.
Based on photos, Robert Groden agreed with the HSCA’s location of McLain. Richard E. Sprague (another photo consultant to the HSCA), however, totally contradicted Groden. Sprague concluded that McLain had not approached the Houston-Elm intersection until well after the last shot. He also strongly doubted that McLain could have made up the necessary ground in the time allotted. In effect, according to Sprague, no motorcycle agreed with the HSCA.
Bugliosi does not discuss the bell tower that Thomas cites (Thomas, p. 643). On the contrary, he quotes Pellicano, who identified a church bell, but this one was not within earshot of Dealey Plaza. The HSCA concluded that the bell sound on Ch-1 did not derive from Dealey Plaza (8 HSCA 112). If Thomas is correct about the bell sound (reportedly heard on November 22, 1964) in Dealey Plaza, I wonder why the HSCA ignored this—or were they unaware of it? Thomas does not comment on this.
Steve Barber claims that Thomas’ key crosstalk (by Fisher, supposedly just seconds before the suspect shots) is actually not crosstalk at all. (See here and here.) Barber agrees that Fisher had said, “I’ll check it,” on Ch-2. On the other hand, on Ch-1, Barber hears something quite different: “Alright, Chaney.” If that is true, Barber argues, then there was no crosstalk at that time and Thomas cannot use the Fisher event as crosstalk—which would seriously damage Thomas’s entire case. As further support for Barber’s conclusion of no Fisher crosstalk, Gary Mack reported that WA had heard this phrase (on Ch-1) as, “Alright, Jackson.”
There is even one more argument against Fisher’s words as crosstalk. On Ch-2, the phrase supposedly is, “I’ll check it.” But that is not the complete phrase. On Ch-2, it actually was, “That’s alright, I’ll check it.” According to Barber, neither Thomas nor Bowles have accounted for the missing words, which really should be on Ch-1 also. Barber concludes: he actually heard a distinctly different phrase on Ch-1 than he heard on Ch-2. If so, Fisher cannot be an example of crosstalk, and that would be devastating to Thomas.
My Summary. Bugliosi (or a ghost) provides powerful evidence from eyewitnesses that McLain’s position did not match the HSCA’s (or Thomas’s) scenario. He even names a policeman who always whistled. Thomas’s purported crosstalk CHECK is not the same sound on the two channels.
The final published critique (from 2005) to be discussed here is by Linsker, Garwin, et al. (hereinafter abbreviated as LG). This report is a sequel to the original NRC report. (Four of the recent authors were also on the original NRC panel.) In this update, they confirm the original NRC conclusion: the suspect shots occur about one minute too late. They also cite the work of Barber and O’Dell, with whom they agree. Their specific methods include cepstral analysis and spectrographic techniques. Although cepstral analysis was originally designed to detect echoes in acoustic signals, it is also useful for analyzing human speech. It is especially helpful for identifying repeats (overdubs). In addition, LG used “pattern cross-correlation” (PCC) to identify crosstalk from two different spectra; these are graphs of power density (raised to the 0.3 power) at each time and frequency. Results of this study can also be used to determine the relative speed of the two channels in question.
One of LG’s chief conclusions is that on the FBI recording of Ch-2 (Track 7—see Appendix 6 for a list of these Tracks)
there can be no doubt that the hum was recorded along with the original sound, and not during any subsequent copying process.
LG claim that this FBI recording is superior to the one (by Bowles) that was used by the original NRC. (Thomas also used the Bowles recording in his analysis.) If this argument (about the hum) is accepted, and it is a very powerful argument, then the FBI recording faithfully represents the sounds from the original Dictabelt (Ch-2), and suspicions about alteration (or loss) of the original can be put aside. It also means that the original (or a direct copy) has actually been used in all of the acoustic analyses to date. Because the speed correction factor is well known (based on the hum frequency), LG’s analysis relies almost exclusively on the FBI copy (Track 7, Ch-2).
They first attempt to synchronize the two channels using HOLD and YOU. They emphasize that this FBI copy (Ch-2) contains no repeats or skips. However, they acknowledge that Thomas disagrees: on the contrary, he reports two utterances on Bowles’s version (Track 2, Ch-2)) that are missing from the FBI copy of Ch-2. For Thomas, this implies forward skips on Ch-2. LG disagree—they claim that there are no forward skips on the FBI copy. Furthermore, the cepstral analysis rules out any backward skips (i.e., repeats or overdubs). LG also tell us something new: the transit time through one groove is 3.65 seconds; therefore when a single overdub or skip occurs, the expected overall time changes by that much (either positively or negatively). It should be added that double skips or double dubs are rare.
If AGC (automatic gain control) had not been used, the recordings could simply be placed side by side and their peaks compared. However, AGC screws things up—it alters the height of the peaks so that comparisons become less reliable. But that is where PCC can help. It measures the power in the spectra; the two channels can then be compared in this fashion and a judgment can be made about time offsets. The relative speed correction can also be varied (in the analysis) in order to locate the strongest peak in the PCC—the best match will presumably identify the proper speed correction. These PCC graphs display the intensity of the sound (raised to the power of 0.3) vs. frequency. To compute the PCC, the power in the two spectrograms (one from each channel) is cross-correlated at each frequency. That correlation will depend on the time shift between channels, but a range of time shifts can be explored. A strong and clear peak in PCC implies that the Ch-1 segment is contained within the Ch-2 segment (at the relative time shift tested); a strong peak is good evidence for crosstalk. These calculations can be repeated for different speed corrections (in Ch-1). The strongest match in PCC identifies the correct speed correction. The position of the peak in time also identifies the time offset between channels. Therefore, PCC determines both the relative timing and speed of the two channels. The authors apply this technique to three proposed cases of crosstalk: CHECK, HOLD, and YOU (using the notation of Thomas).
Cepstral analysis employs a time-varying signal input, with some repeat or added components to the signal (possibly with attenuation), including some time delays; this is reminiscent of the echo patterns in Dealey Plaza. Cepstral analysis is also useful for identifying stylus jumps during playback of the Audograph disc.
The authors also use a Gabor spectrogram (probably named after Dennis, not Zsa Zsa—although both were Hungarians). This approach uses specific filters, so that the resolution of the signal power, in both frequency and time, can be optimized.
The authors then begin to synchronize two crosstalk events: “...HOLD everything secure...” (by Decker) and “YOU want me to...” (by Bellah). If these are authentic crosstalk events, then synchrony should be evident. Thomas had suggested (they cite him for this) the presence of forward skips on the FBI recording for Ch-2 (Track 7). LG disagree—they find no evidence of a forward skip. Furthermore, by means of cepstral analysis they also conclude that there are no repeats (backward skips). On Bowles’s version of Ch-2 (Track 2), however, they find two repeats followed by a forward skip. Aside from that, in this time interval, the Bowles version (Track 2) and the FBI copy (Track 7) match just fine on Ch-2 for every utterance.
LG then focus on Thomas’s most critical evidence for crosstalk, abbreviated here as CHECK: “I’ll check it...” (by Fisher, according to Thomas). Thomas had located these words just seconds before the suspect shots on Ch-1. Because of this close proximity in time, he regarded this event as highly reliable for achieving correlation between Ch-1 and Ch-2. (Thomas had not, however, discussed Fisher as a tie-in during his 2001 article; he only introduced it later.) LG first emphasize that if CHECK had been authentic, it would be incompatible with both HOLD and with YOU (which LG accept as valid crosstalk events). Furthermore, this discrepancy cannot be due to lost time on Ch-1, since there was none—the constant motorcycle noise prevented that. In addition, their PCC analysis showed no prominent peak for CHECK, which also argues against this event as a crosstalk. On the other hand, if there had been two different utterances of CHECK (at two different times) then this peak should indeed be absent, as it was. In the case of YOU and HOLD, LG’s analysis strongly confirmed true crosstalk, but the opposite was found for CHECK—i.e., these two cases of CHECK were likely spoken at two separate times and therefore could not represent crosstalk. If this conclusion is accepted, Thomas’s newly applied tie-in (via Fisher) is mortally wounded.
LG (p. 222 in their article) next analyze the time interval between GO (Curry’s phrase on Ch-2, which is not an example of crosstalk) and the suspect shots (Ch-1), by using the crosstalk HOLD. On the FBI copy of Ch-2 (Track 7) the time interval between GO and HOLD is 67.7 seconds; on Bowles’s version of Ch-2 (Track 2) it is 64.3 seconds. Since LG claim no skips or repeats during this interval they conclude that the actual elapsed time is 67.7 seconds (the FBI copy was deemed more reliable). Next, they state that the time from the first suspect shot to HOLD was 6.6 seconds (from their Table 1). Since Ch-1 had no down time (the motorcycle kept the recorder going), they conclude that 6.6 seconds is the actual elapsed time. They therefore conclude that the time from GO (which came first) to the first suspect shot was 61.1 seconds (67.7-6.6 = 61.1). We have already seen this argument from O’Dell. Keep this time interval in mind as the discussion proceeds.
Another supporting argument for the validity of HOLD (as a valid crosstalk) is the bell sound. This occurs 7.7 seconds after HOLD on the FBI’s Ch-2 (Track 7) and 7.8 seconds after HOLD on Bowles’s version of Ch-1 (Track 1). That agreement is quite impressive.
If crosstalk YOU is employed instead, then on Ch-2 the time interval from GO to YOU (on Track 7) is 67.6 + 143.2 = 210.8 seconds. That same interval, on Bowles’s version of Ch-2 (Tracks 2 and Track 3) is 63.9 + 148.4 = 212.3 seconds. On the other hand, the Ch-1 time interval from the first suspect shot to YOU (Track 1) should be longer than this—because the starting point (the first suspect shot) should logically occur before the GO utterance (which was a response to the shooting).
However, the apparent time interval (on Bowles’s Ch-1, i.e., Track 1) from the first suspect shot to YOU is actually shorter: 6.5 +173.0 = 179.5 seconds. That same interval on the FBI’s copy of Ch-1 (Track 5) is 368.7 -187.7 = 181.0 seconds. Combining these results yields the actual elapsed time from GO to the first suspect shot: 210.8 – 179.5 = 31.3 ± 1.5 seconds (plus the Ch-2 dead time during the interval, GO to YOU). This, of course, clearly implies an illogical sequence, namely that GO (an obvious response to the shooting) occurred before the suspect sounds—but that is precisely what the arithmetic requires. The dead time in question consists of five periods of radio silence, supposedly at least four seconds each; this should add at least another 20 seconds, which would increase the total interval (from GO to the suspect shots) to at least 31.3 + 20 = 51.3 seconds (but likely somewhat more). Therefore, the agreement with the time interval calculated above (the number the reader was asked to keep in mind—61.1 seconds) is really quite good. If this conclusion is accepted, all of the acoustic probability calculations by the HSCA (and by Thomas, too) become totally irrelevant—the suspect sounds simply occur at the wrong time to be real shots.
LG emphasize that Thomas primarily employed the NRC report and the Bowles recordings (Track 2) from the Gray Audograph (Ch-2). This latter copy contains many repeats; the NRC considered it less reliable than the FBI copy (Track 7, which LG chiefly used). They next cite Thomas for four errors, as follows:
On Ch-1 (Track 1), the playback time between HOLD and YOU is 172.9 seconds. (For these time intervals, see Table 1 in LG’s article, i.e., my Appendix 4, reference 6.) Therefore the actual time between HOLD and YOU is (172.9 – SB) seconds, as shown in the sketch. (If skipback had occurred, as Thomas proposes for HOLD, then that utterance would appear too early on the tape and the resulting interval, from HOLD to YOU, would end up being longer than it actually was, because of that head start. Therefore, to arrive at the real time interval, some time (SB) must be subtracted.) On Ch-2 (Track 7), the playback time interval, from HOLD to YOU, is 143.1 seconds. So the actual time on Ch-2 from HOLD to YOU is (143.1 + DTHY) seconds. These two expressions should be equal, which yields:
172.9 – SB = 143.1 + DTHY, which leads to
SB = 172.9 – 143.1 – DTHY = (29.8 – DTHY) seconds.
Since this latter expression must always be positive or zero (time intervals can never be negative), then DTHY can never be greater than 29.8 seconds:
DTHY (dead time between HOLD and YOU) < 29.8 seconds, and
SB < 29.8 seconds (SB must be less than or equal to 29.8 seconds).
Next LG look at the time interval between CHECK and HOLD. They continue to use Thomas’s scenario, which assumes that CHECK and CHECK1 represent a true crosstalk event. (Refer to the above sketch.) This is 12.5 seconds (the playback time as measured on Ch-1); to obtain the actual time, however, we must now add SB, so we get (12.5 + SB). (Due the assumed skipback, this interval would have appeared shorter on playback, by just as much time as the interval from HOLD to YOU would have appeared longer during playback. Therefore SB must be added to obtain the actual time.) On Ch-2 (Track 7) the playback time is 99.1 seconds. Therefore the actual time from CHECK to HOLD is (99.1 + DTCH). As before, we can equate these two expressions, since they describe the same time interval:
12.5 + SB = 99.1 + DTCH, or
SB = 86.6 + DTCH.
Since DTCH represents real time it must always be positive or zero. Therefore,
SB > 86.6 seconds (SB must be equal to or greater than 86.6 seconds).
But this disagrees with the prior result, where SB had to be less than 29.8 seconds! Because the above two conclusions for SB disagree (by almost a minute: 86.6 – 29.8 = 56.8 seconds), the assumptions in the calculations cannot all be true. So, using Thomas’s scenario, LG have arrived at their reductio ad absurdum. One of Thomas’s assumptions is that CHECK is a valid tie, so that becomes suspect. LG finally conclude that (a) CHECK cannot be a valid crosstalk, (b) there is no evidence for a skipback (SB) on Ch-1, and (c) YOU is a valid crosstalk.
LG add that there is even more evidence for HOLD as a valid tie: the presence of strong heterodynes (heard as beeps) at this same time strongly imply that the crosstalk sounds had actually been recorded live at the moment of the crosstalk—and had not been superimposed at a later date. They also emphasize that the HOLD crosstalk is further validated by the bell sounds—the intervals are consistent on the two channels. Therefore, if HOLD is a valid crosstalk (recall that it overlaps the suspect sounds), then the first suspect shot occurred 61 seconds after Curry’s “Go to the hospital”—and the acoustic case for conspiracy is dead.
In their final letter (Appendix 4, citation 8, Figure 2), LG add two fascinating images related to CHECK and CHECK1. They display the spectrogram from the FBI copy of Ch-2 (Track 7) for “I’ll check it.” This is like a voiceprint, vaguely similar to a fingerprint. Their next figure (Figure 3) does the same for Thomas’s purported crosstalk CHECK1, from Ch-1 (Track 1). Whereas we learned above that several auditors heard something other than “I’ll check it” at this point on Ch-1, with these two new images we can now see how different these voiceprints are. In other words, our ears and eyes both agree that CHECK cannot represent crosstalk.
To conclude, the authors admit that they have ignored the statistical argument (as developed by the HSCA, and later by Thomas), insisting that the suspect sounds were irrelevant, because they came too late. To put the matter most simply: once HOLD is accepted as a valid crosstalk (i.e., no skipback on Ch-1), then no further timing analysis is required. The suspect sounds then overlap with “...Hold everything secure...”. Since, by the simple logic of human events, this phrase must have occurred after the shooting (not during the shooting) then the suspect sounds cannot represent real shots—and the acoustic case for conspiracy is dead.
My summary. LG’s scientific analysis is very persuasive. Furthermore, it matches the eyewitness evidence. If CHECK and CHECK1 are assumed to be crosstalk, a deadly paradox ensues in the timeline. Moreover, LG assemble multiple other lines of evidence against these utterances as crosstalk, including their voiceprint images. If CHECK is thrown out, Thomas’s case is history. Finally, the scientific evidence for both HOLD and YOU as crosstalk events looks sturdy. In particular, if HOLD (“...Hold everything secure...” by Decker) is accepted, then its utterance is concurrent with the suspect sounds, the Double Decker wins, and the case is closed.
Despite the earlier worries that the Dictabelt (at NARA) might not be an original, the extant Dictabelt No. 10 (at NARA) is most likely the original (based on LG’s hum analysis). Optical scanning (i.e., digitizing at the Lawrence Livermore Laboratory in Berkeley) might help to clarify issues even more (especially with respect to the hums), but six years have passed since that option was first raised, and we have heard nothing further from NARA. I suspect that money is in short supply. (Wealthy donors are welcome.) In view of the uncertain chain of custody for the Dictabelts, a clever lawyer might still exclude them from the courtroom—unless, of course, the work of LG (regarding the hums) was persuasive to the judge. (There are at least two more items that might be banned from the courtroom: the Z-film and the magic bullet. Even the skull X-rays might find themselves as outcasts—I would agree to testify against them.)
As Thomas emphasizes, the order in the gunshot data is indeed impressive, and it is also consistent with the spacing on the street. But the suspect sounds might derive from other sources. Also recall that for a given microphone, the echo patterns were surprisingly similar for shot origins that were far apart in space.
Recall that the WA analysis was strictly virtual—no data were ever taken to confirm their calculations. And no one has ever tried to duplicate this—not even by computation. That their match was so extraordinary for the suspect GK shot (26 of 26 Dictabelt impulses matched to the multiple impulses for the test shot) is impressive, although the NRC did try to whittle this down. In fact, they eventually got the probability into the range of insignificance. However, it is disappointing that not a single calculated impulse pattern by WA was ever verified by experiment. To check every one would have been overwhelming, but to check a few should have been feasible. And what if other Dictabelts (even from November 22) had been similarly analyzed: Might they also have shown such remarkable matches? (NARA still has them.)
Sergeant Bowles told Harry Livingstone that, several days after the assassination, he had taken tapes to Oklahoma, where they were copied. What happened there is not known. In addition, Bowles added that they had essentially worn out the belt and disc by repeated playing and that this may have caused some of the spikes (Killing the Truth 1993, Harrison Livingstone, pp. 353-354).
Bowles also told Livingstone that it was rather common for police recordings (on other occasions) to record gunshots (which were confirmed as gunshots by officers at the shooting site)—and that these were easily recognized as shots on replaying. On the other hand, on the Dictabelt for this case, Bowles did not recognize any gunshots. As WA stated, these were not actually inaudible, but rather just sounded like static to the human ear. It is unfortunate that some of these other Dictabelts (see Table 16 by Thomas), even for November 22, were not subjected to a similar analysis—just to see how many (if any) false matches existed. So far as I know, only Dictabelt No. 10 was scanned for such matches, but none were found on that one. (Well, there was one exception discussed above—the so-called 6th shot—but this did not satisfy the screening criteria.)
McLain’s disagreement with the HSCA (and with Thomas, too) about his location is serious. Why did he change his story? His answer is that he felt pressure during his initial interview in Washington, mostly because he had been asked misleading questions, so afterwards he just wanted to clarify his recollections. He also emphasized that he had not listened to the tapes until after his HSCA appearance, which seems an odd method for the HSCA to do business. After this entire review, I am strongly inclined to believe McLain—i.e., he did not have an open microphone.
No crowd noises or sirens appear on Ch-1, even though the stuck microphone should have picked them up. By contrast, they were easily heard on Ch-2. Even whistling can be heard (twice) on Ch-1. The bell, too, was easily heard. So why were the gunshots so hard to hear on Ch-1?
To my surprise, Thomas does not mention silencers, which were available in 1963 (See “Silencers, Sniper Rifles & the CIA,” Probe, Nov-Dec 1995, Carol Hewett). It is critical to recall that silencers can only effect the muzzle blast. They cannot affect the sonic wave, since that is produced during the flight of the bullet, far away from the gun. Did any Dictabelt patterns (of sonic waves) suggest silencers? No test shots were fired with silencers so here is another black hole.
As a believer in a JFK conspiracy, I began this review hopeful that Thomas was right. But I have become more and more dubious about the acoustics case for conspiracy. The arguments advanced by LG (and O’Dell, too) are particularly telling. (But recall that it was Barber who triggered this skeptical chain of events.) It is unfortunate that more data were not taken, e.g., at least some data to confirm WA’s virtual simulations; even a limited sampling might have helped. Most researchers would also like data for other shooting sites. It cannot be overemphasized that the HSCA used only two: the TSBD and the GK. The storm drain (on the overpass) would have been interesting. So, in the end, the acoustics data tells us nothing about other possible shooting sites.
Mostly in this review, I have ignored the issue of Z-film authenticity. Although it is a fundamental disagreement between Thomas and me, it was not necessary for me to invoke those beliefs in order to doubt the acoustics case. However, if the film has been altered, then Thomas’s remarkable audio-visual agreement for the critical time interval between his first (Z-224) and last (Z-313) shots, is probably bogus. These problems of Z-film authenticity were not acknowledged by him. Although he admits that the audio-visual correlation cannot be closer than five Z-frames, we already know that a prior correlation (by the HSCA) was quite different from his—in fact, they disagreed with him by much more than five frames. Furthermore, we have seen how even BBN often disagreed with WA by more than five frames. As one more example, Alvarez changed his mind by up to 20 frame intervals (!), based on evidence that was supposedly objective. Therefore, the close correlation (of five frames) that Thomas claims for his audio-visual data may just be Madame Luck smiling at him.
If I were to base my entire case for conspiracy on the acoustics data I would feel very insecure. On the other hand, the primary evidence for conspiracy in this JFK case (i.e., the medical evidence and the LHO items) is so compelling that the acoustics data might then serve as a light auditory encore. Unfortunately, that melody seems to have ended on a discordant note.
In spite of this, though, I still admire Thomas’s stamina and courage for so meticulously picking his way through this massive and chaotic knoll of evidence. As a solitary prophet in this noisy, echo-filled wilderness, he has kept this peculiar story in play for a long while. For that we all owe him and wish him well. However, many doubting Thomases likely still have their guns cocked for Thomas. All that is missing are the sounds of gunfire. No doubt, though, Thomas has learned to bob and weave, and will not easily exhibit a (metaphorical) head snap.
Will Thomas be disappointed that I became unconvinced? I don’t think so. After all, he is the outlier—the conspiracy believer who accepts the SBT. That’s about as paradoxical as it gets. So why not a conspiracy believer (like me), who is not convinced by the acoustic data? To paraphrase a chief suspect in this JFK case (he was a relative of a former patient, who gave me a family photo), this house has many rooms. James Jesus Angleton, of the CIA, had offered precisely that comment when asked who had killed JFK.
So, if the suspect sounds were really not shots, then what were they? Bowles thought that impressions in the disc from frequent replaying were responsible. O’Dell thought that normal speech patterns from hard consonants caused (or contributed to) the suspect sounds. Or perhaps they came from the motorcycle ignition system, or from the engine itself. WA suggested another possibility: electrical or mechanical disturbances in the DPD radio transmission, reception, or recording equipment. McLain even said that these random radio noises could sound like gunshots. WA stated: “Some test more discerning than the human ear was required to determine the probable cause of the sound impulses.” Perhaps the actual source of the suspect shots was not haphazard, but rather had a persistent and genuine physical basis (as O’Dell speculated). If so, then BBN should not have compared the suspect sounds to randomness, but instead to a real physical source. Unfortunately, the HSCA never seriously considered such an option, so an adequate explanation has evaded us. On these grounds, as well as many others, I suspect that this discussion shall continue. Perhaps Thomas still has more surprises for us.
FLASH! This book is now available on KIndle for the lowest price ever, of $10.99
New Edition, Updated!
Destiny Betrayed: JFK, Cuba, and
the Garrison Case by James DiEugenio
Order Now! Amazon.com
Enemy of the Truth: Myths, Forensics
and the Kennedy Assassination
by Sherry G. Fiester
Forensics can be a complicated subject, yet Fiester provides the reader with easily understood, accurate, information. Enemy of the Truth: Myths, Forensics and the Kennedy Assassination is so comprehensive in its approach, this work should be used in the instruction of all new crime scene investigators nationwide. William LeBlanc, CFCSI