CRIMINAL ACOUSTICS:

 

IBM Launches Counter-Attack on the JFK Evidence

 

 

The recent issue of  Science & Justice (vol. 45: 207-226) features an article authored by Ralph Linsker, Richard Garwin, Herman Chernoff, Paul Horowitz and Norman Ramsey, (NRC redux) which asserts that the acoustical evidence of gunfire in the Kennedy assassination, developed by the House Select Committee on Assassinations in 1978, is invalid. In making this assertion the authors do not address the core acoustical evidence. Rather, as in their earlier discredited* study, many of the same authors (Ramsey et al. 1982) reassert their opinion that the putative gunshot sounds were not recorded simultaneous with the assassination. To arrive at this conclusion the authors of the article do not argue from the evidence as much as they argue around the evidence. Within the scope of their study, the methods and resulting data are entirely sensible. The results of the new study corroborate the timeline calculations produced by Michael O’Dell (2003). Even more importantly, the new voice-print study strongly corroborates the crosstalk evidence that the putative gunshots are simultaneous with the shooting. There is a striking disconnection between the IBM results and their conclusions.

 

The objective of the IBM study, aside from rebutting my article (Thomas 2001), was to establish an accurate timeline of the sequence of events as they were recorded on the 22nd of November 1963. The problem is much more complicated than one might have wished and these complications provide the wiggle room, or perhaps more appropriately, the cover, for the disconnect between the IBM results and the IBM conclusions. The original DPD recordings are studded with anomalies caused mainly by the use-worn recording and playback equipment, much of it inflicted during the Warren Commission era in the making of transcripts and copies. Perhaps one of the most bizarre, but fortuitous, anomalies is the fact that the radio microphone on a police motorcycle became stuck open for about 5-1/2 min, overlapping the time of the assassination. It was in the midst of this recorded motorcycle noise that the HSCA acoustical experts discovered the assassination gunfire. Most of the technical work by the IBM lab is an effort to sort out the various anomalies, especially those on the Channel 2 recording, which contains the broadcasts from the President’s motorcade. The police were broadcasting over two radio channels that day. The primary channel used for routine communications, Channel-1, was recorded on a Dictaphone belt recorder. An auxiliary channel, Ch-2, used on this day for communications from the police escort of the Presidents motorcade, was recorded on a Grey Audograph disc machine. Both machines were stylus-in-groove arrangements, and both used the same stylus which cut the original acoustical groove to playback the recording, and consequently, both suffered considerable attrition due to multiple playbacks for which purpose they were not designed. Importantly, the police made magnetic tape copies of the recordings before they became completely degraded. But even these taped copies, especially the Ch-2 recording, have multiple skips as the stylus displaced during playback of the worn disc. The skips are both forward, causing missing ____________________________________

*Their timelines were shown to be wrong by O’Dell (2003); their statistical analysis was shown to be wrong by Thomas (2001).

segments, and backward, causing repeated segments. The skips are especially obvious on Ch-2 because it is dense with voice broadcasts, whereas the corresponding period on the Ch-1 recording is mainly motorcycle motor noise. With respect to constructing a timeline of events, a further challenge was to match playback speed of the tapes to the original recording speed. Neither machine was a precision chronometric device, even when new. JC Bowles states that both machines were prone to noticeable changes in speed during operation. Moreover, because they were designed for dictation, Dictaphone machines had a speed adjustment by means of a knob on the front of the machine, which the operator could set to preference. Hence, playback of a dictabelt on a machine different from the one which made the recording, or for that matter, on the same machine at a later time, would inevitably be different from the original recording speed. A primary objective of the IBM study was to resolve the difference between playback time and the original recording time. Their results are in accord, within a percentage or two, of the times reported by Michael O’Dell (2003). But the IBM study also demonstrated that there are limits to the accuracy that is attainable by any attempt to measure the amount of time that was recorded, and thus, the amount of time that was not recorded. Recorded time is not actual time. The recorders were designed to stop whenever there was dead-air. Unfortunately, there is no direct way to detect recorder stoppage, nor the amount of time that the recorder was stopped, if and when it did stop. Similarly, skips are easily detectable when they occur during broadcasts of speech, but not so easily detectable during periods of silence or radio noise. Rather, such impositions, and the time intervals involved, must be inferred indirectly by measuring the time intervals between events that are common to both channels. There is an array of simulcasts that are found on the recordings, and in every case, the amount of recorded time between the simulcasts is different on the two channels. Recorder stoppage by itself cannot account for all of these differences. Because this data contradicts an assumption key to their conclusion, The IBM report does not include this information in their analysis. The data is provided here in Table 1.

 

 

Table 1.- Speed corrected playback time intervals between simulcasts on the Dallas

Police recordings (in seconds). Data from O’Dell (2003)

 

                                    INTERVAL                               CH-1         CH-2             Δ

 

                               CHECK to HOLD                            10              99               89

                                 HOLD to YOU                             174            143              31

                                   YOU  to ALL                               15              12                3

                             YOU to ATTENTION                     114              90              24

 

The IBM and NRC conclusion that the putative gunshots are not synchronous with the shooting rests on the juxtaposition of the Decker broadcast, an accidental simulcast of the words, “Hold everything secure.” Decker was transmitting over Ch-2, but the words cross-talked on to Ch-1. His original transmission occurred more than a minute after the shooting. But on Ch-1 the words are virtually simultaneous with the last of the sounds identified acoustically as one of the gunshots. Logic dictates, and the IBM report concludes, that the juxtaposition of the two sounds proves that the suspect sounds must have been recorded long after the assassination and therefore cannot be the assassination gunfire. Alternatively, James Barger, the lead scientist on the original acoustics study, suggested in a letter to the chief counsel for the House Select Committee on Assassinations that the apparent juxtaposition of the two sounds could be an artifact caused by the imposition of a stylus displacement during the recording phase, or, by an overdub during the copying phase. Because there is no evidence for either scenario the suggestions are no more than conjecture. However, now that the existence of multiple cross-talks are known, and because it is further known that in every instance the time intervals between them is different, then the time offsets are a fact, not conjecture. The precise mechanism causing the offsets in any given instance remains conjectural. But whatever the specific cause, the offsets prove that the apparent juxtaposition of events on playback does not reliably establish the true time interval between events.

 

Failing to acknowledge the simulcast data is not the only error of omission committed by the IBM group. In spite of their claim that the putative gunshots on Ch-1 are not synchronous with the shooting, they never actually fix the time of the assassination. This omission prevents the reader (and the editor of the journal) from seeing the contradiction in their arguments. The time of the assassination can be fixed by two key broadcasts over Ch-2 originating with Police Chief Jesse Curry who was in the lead car of the motorcade. Prior to the assassination Curry had been broadcasting the location of the motorcade at regular intervals as it wound through the streets of Dallas. The first broadcast indicating that the shooting had occurred was a shouted order from Curry to the police escort to “Go to the Hospital officers, Parkland Hospital.”  Twenty seconds earlier (recording time) Curry had stated his location, “…at the Triple Underpass.” The first word of this particular broadcast is garbled. A transcript made by the Dallas Police in 1963 when the recordings were relatively new reads, “approaching the Triple Underpass.” Whether at or approaching the underpass, the broadcast places the lead car on Elm street at the west end of Dealey Plaza, the scene of the assassination. The President’s limousine was about 150 ft behind the lead car, as shown in a photograph of the two cars in Dealey Plaza taken by Jack Weaver (FBI exhibit Q220).*

Thus, when the lead car was at or near the underpass, the President was in the kill zone in the mid-section of Elm Street, and therefore, the broadcast by Curry “at the Triple Underpass” must have been made within moments of the assassination. The distance from the Elm-Houston intersection to the Triple Underpass is 500 ft. The Warren Commission calculated that the first shot was

  __________________________________

*The picture was published on the jacket of Six Seconds in Dallas by Josiah Thompson. The two cars are also briefly visible at the end of the Zapruder film. The lead car stopped in the underpass and waited for the President’s limo to catch up, probably in an effort to establish the President’s status.

fired when the limousine was approx. 150 ft west of the intersection (based on the Willis photo). The Warren Commission also calculated that the entire shooting occurred within 7.9 sec (WR pp. 111-117). The acoustically identified gunshots occur within an interval of 8.3 sec. Thus, Curry’s critical broadcast over Ch-2 serves as a real-time marker for the time of the assassination.  

 

The broadcast which immediately precedes the Triple Underpass announcement, by just two seconds, is a transmission from Deputy Chief Fisher saying, “Naw that’s all right, I’ll check it.” The latter words, “I’ll check it” are also present on Ch-1. In a transcript prepared by JC Bowles, supervisor of the DPD communications department in 1963, this broadcast is cited as the exemplar of the crosstalk phenomenon, an accidental simulcast. The broadcast of “I’ll check it” on Ch-1 occurs just two seconds before the first acoustically identified gunshot sound. Thus, the Fisher crosstalk establishes exact synchrony between the acoustically identified gunshot sounds and the actual shooting. This inconvenient fact is never mentioned in the IBM report, lest they have to explain the diabolical coincidence established by the Fisher crosstalk.

 

Because in correspondence I had pointed out the significance of the Fisher simulcast, The IBM authors were compelled to challenge the interpretation of the “I’ll check it” broadcasts as actual crosstalk. At my suggestion they performed a voice-print analysis, a comparison of the frequency spectrum of the two utterances. They also applied the same test to two other instances of crosstalk that occur between one and two minutes later. One was a broadcast by Sheriff Bill Decker saying, “Hold everything secure.” The other was a broadcast by Sergeant Bellah saying, “You want me to hold traffic on Stemmons.” Both broadcasts originated on Ch-2 and crossed over onto Ch-1. The comparison of the

corresponding broadcasts is made by a computer which slides a two sec segment containing the utterance from Ch-2 against a ten sec segment of Ch-1 containing the corresponding utterance, seeking a match. The result of the comparison is scored mathematically as a moving correlation coefficient, which is printed out as a graph. A peak in the graph at the correct delay (the position of the utterance) indicates that the computer found a match, that is, a significant similarity in the frequency spectrum of the broadcasts.

 

Either side of the peak in the graph is the background level where there is only accidental but insignificant similarity, usually less than ten percent. In all three comparisons, including the Fisher crosstalk, the computer produced a peak. Because this result contradicted their assertion that the “I’ll check it” broadcast is not actual crosstalk the IBM report does not include this graph. Because the graph is withheld we also do not know the amplitude of the peak. The IBM report only reveals in the text that the peak obtained was “no larger than many accidental peaks.” They do provide a graph showing what they consider to be accidental peaks generated by the computer comparison. Moreover, some of these accidental peaks are of the same amplitude as the peaks obtained with the cross-talks favored by them as real crosstalk. The unstated implication of their dismissal of this evidence is that the peak obtained by the computer analysis was also accidental and just another diabolical coincidence. In my opinion, objective analysts would have produced the results of their test regardless of the outcome, and if their convictions were honestly held, would have stated their belief that the result was a false positive. Neither is found in the IBM report. Instead they dismiss the result for three stated reasons. Firstly, they argue (on p. 221) that the broadcast on Ch-2 cannot be a crosstalk from Ch-1 because,

 

“if CHECK were a valid crosstalk its timing would be incompatible not only with HOLD, but also with the timing of the well established crosstalk YOU…”

 

The references HOLD and YOU are to respectively the Decker and Bellah crosstalks listed in Table 1. They refer to the Fisher pair as, CHECK. Of course, their argument is a non-sequitur. The incompatibility is with their conclusions, not with the evidence. The data in Table 1 shows that because of the offsets all of the cross-talks are “incompatible” with one another. This probably explains why the crosstalk data does not appear in the IBM report. A reader armed with the information in Table 1 would be likely to conclude that the apparent lack of synchronization could be explained by the time offsets caused by the recorder anomalies rather than by an incompatibility in real time. Furthermore, the greater the separation of events in real time, the more likely it is that an imposition will have offset their juxtaposition on the recordings.

 

Secondly, the IBM report argues that the PCC peak obtained was not at the expected speed warp. But a comparison at the incorrect speed would not generate a peak. A comparison at the wrong speed can generate a false negative, but it cannot cause a false positive. A peak can only indicate that the frequency spectrum is similar to a significant degree. The accidental similarity is manifest in the baseline of the graphs, not the peaks. More importantly, the expected speed warp is meaningless in this context. The expected speed warp is determined from the 60 Hz motor hum. But it is known that the actual speed of the recording mechanism at any given instant wobbles around the motor speed. It is precisely for this reason that the computer makes the comparison iteratively at increments of speed warp deviating from the expected speed. In a classic example of circular reasoning, when the test was applied to the crosstalk instances favored by them, the “correct” speed was defined as the speed that generated a peak!

 

Actually, the PCC method is not definitive. That is proven, for one thing, by the so-called accidental peaks (if they really are accidental and not causative; we don’t know what caused the accidental peaks). Secondly, the test cannot prove identity between the signals but only significant similarity. By definition, if two signals are perfectly identical they would have a PCC score of 1.0. The HOLD and YOU cross-talks only scored around 0.3 against an “accidental” baseline of 0.1. The explanation for the lack of identity is because the noisy backgrounds are different on the two channels and the noise to signal ratio is high. Hence a peak only proves significant similarity. In the IBM report they suggest that the peak obtained in the test of the CHECK broadcast might have been caused by the Ch-1 utterance being similar, by coincidence, to the utterance on Ch-2. They demonstrated this alternative by comparison of two separate broadcasts of the words “ten-four” by the radio dispatcher. The computer comparison generated a strong peak. Such a coincidence, which in this context would be a false positive, could explain the peak obtained in the comparison of the CHECK broadcasts. Perhaps someone else with a vocal quality similar to Fisher’s said the words “I’ll check it.”!

 

Relevantly, IBM had a test which can separate a false positive peak, from a true positive. They called this test the d-warp test. The “d” stands for duration. The peak obtained in the comparison of the dispatcher’s broadcasts would be expected because the words and vocal quality measured in the frequency spectrum, are the same. But the duration of the transmission would be slightly different just because they are different utterances, whereas the duration would be exactly the same if they were true crosstalk. The IBM group applied the d-warp test to the HOLD and CHECK broadcasts demonstrating that at the speed warp which generated the peak, the duration was also the same. Conversely, at the speed warp which generated the peak with the dispatchers “ten-four” broadcasts the duration was different, demonstrating the principle and the power of the test. The IBM report states (on p. 222) that the peak from the comparison of the CHECK broadcasts had failed the d-warp test. But again, the results were withheld. Only on closer examination does it become clear that the d-warp test was applied not to the peak obtained at the correct delay (but “unexpected” speed warp), but to one of the accidental peaks obtained at the expected speed warp, but not at the expected delay. By this sleight of hand they seek to convince the reader that the CHECK broadcast is not a crosstalk. One might suppose that the reports authors really did apply the d-warp to the peak at the correct delay, as they certainly should have, but the result is not in their article. It is difficult for this reader to believe that they would have withheld a result which supported their conclusion.

 

A minor issue, but one which is instructive, is the claim in the IBM report that my article (Thomas 2001) contains serious errors. But most of the “errors” that they cite are not in my article but in my correspondence with them, nor are they errors of fact, but rather matters of interpretation. Apparently having an opinion differing from theirs is to fall into error. For example, the IBM report attributes the identification of the Fisher crosstalk to me, rather than to JC Bowles, and then cites this identification as a “serious” error on my part. The purpose in doing so was to hide the fact that the identification of this crosstalk was made by objective, knowledgeable persons, as opposed to an obviously biased individual, me. It also enabled them to attach another “error” to my name in an effort to discredit my criticisms of their work.

 

The only material error in my article resulted from my reliance on their faulty calculations. The IBM report (on p. 226) is correct in stating that I relied on the erroneous parts of their original report in challenging their logic that the acoustically identified gunshot sounds were not deposited on the recordings at the time of the shooting. The implication is that there are parts of the NRC report which are not erroneous. But on all substantive issues the NRC report is substantially wrong. This instance serves as another example. One of the key events on the recordings is a broadcast by Sgt. Bellah over Ch-2 saying, “You want me to hold traffic on Stemmons?” His words cross-talked over to Ch-1 and serve therefore as a tie-point between the channels. On Ch-1 the time interval between Bellah’s broadcast and the putative gunshots is 179 sec. On Ch-2, the time interval between Bellah’s broadcast and Curry’s broadcast, “Go to the Hospital” is 197 sec. However, the playback contains a number of skips and repeats, audible as repeated broadcasts (skips backward) and interrupted broadcasts (skips forward). The NRC panel corrected for the skips and calculated the actual time interval to be 180 sec.  Therefore, my article contained an “if-then”statement on (p. 29),

 

 “…if one uses the Bellah cross-talk to synchronize the transmissions of the

 two police channels, instead of the Decker calls, then the putative gunshots

exactly overlap the interval defined by Chief Curry’s two broadcasts and

occur at the exact instant that John F. Kennedy was assassinated.”

 

I was wrong, because the NRC calculations were wrong. We now know, thanks to Michael O’Dell, who obtained an Audograph disc machine and studied its function, that the machine had a fixed stylus, not a floating stylus like a phonograph machine. Rather, the turntable is driven across the stylus perpendicular to the stylus arm by a worm drive mechanism. Hence, while the stylus had enough play that it could skip, it could not accumulate skips. Rather it would compensate for skips in any one direction by a counter skip in the opposite direction. Thus, the corrections made by the NRC panel were unnecessary. 

 

The NRC panel did realize that there was a strong possibility that there were undetected skips. What was needed was some way to play the disc without stylus displacements. Therefore, with the help of the FBI lab, they rigged a phonograph machine to play back the disc. The problem with using a phonograph machine is that it had to be rigged to play the disc from the inside out, as the Audograph machine does. Secondly, a speed warp was introduced because a phonograph machine plays at steady revolutions per minute, whereas, the Audograph machine plays at track speed, which is inches per minute. Thus, when measuring time intervals on the phonograph playback a correction has to be made for the introduced speed warp. The NRC panel calculated that the time interval from Bellah’s YOU to Curry’s GO to be 206 sec. The newest measurement is 210 sec (by O’Dell) or 211 sec (by IBM). They are in close agreement because they apply the same methods to the same recording.

 

The fundamental question, however, is why should one accept the timelines from the phonograph playback, with its built in speed warp, as more accurate than the Audograph disc playback. Both the NRC panel (on p. 67) and the NRC redux (p. 215) claim that the phonograph playback is free of skips.  That would be a good reason if it were true, but it is not.

 

As a result of my criticism of them, the NRC was induced to release the phonograph playback into the public domain. On listening to the playback the skips are immediately obvious. The recording begins with an obvious repeat. In my correspondence with them I pointed out that there are also five broadcasts found on the Audograph playback that are absent from the FBI’s phonograph playback. In response to this information the IBM lab timed the missing segments and were able to demonstrate that two of the missing transmissions were actually caused not by skips, but by a volume failure during the FBI’s playback. Another missing transmission was not a broadcast at all but a garbled transmission from a skip in the police playback that freakily reproduced as the audible word “twenty.” But the repeat and two skips (missing broadcasts accompanied by a time gap) remain. The IBM report falsely states (on p. 215) that I told them of two skips. I told them of five missing broadcasts, one repeat, and one additional time offset. The IBM report attempts to define the problem away by narrowing their analysis to the portion of the playback which they call “track 7.” The skipped transmissions and repeat transmission occur previous to this segment. The problem however, is that these aforementioned skips are detectable because they occur during segments of the recording where there are voiced transmissions. What of skips during radio noise or silence? It was precisely because of this problem that the NRC attempt at correcting for repeats during the interval from GO to YOU failed. To address this problem the IBM lab applied a method called cepstral analysis that could seek out repeat segments of noise. But contrary to their claim that this analysis revealed no such segments, it did find several. But these were explained away by them, without proof, that they were probable instances of the “premonitory whispers,” another anomaly that plagues these recordings. One is compelled to ask, what is the point of doing such an analysis if one is simply going to ignore the inconvenient results?

 

Another artifact that I pointed out to them which is also evidence for a missing segment is  the offset in the interval between the two Bellah cross-talks. The interval between the broadcasts is 15 sec on Ch-1, but only 12 sec on Ch-2. Because the rotation time of the Audograph disc is about 3 sec, the simplest explanation for the missing time is a skip in the playback. Alternatively, the Ch-2 recorder may have stopped for 3 sec, or there might have been an imposition on Ch-1 that added 3 sec. There is no mention of this anomaly in the IBM article, or of the repeat broadcast or of the two skipped broadcasts. Had they mentioned those anomalies it would tend to dilute their citation of my challenge to the integrity of the FBI playback as a serious error on my part. They do state correctly that if forward skips do exist the amount of time is small, and correcting for them would actually add time to the Ch-2 timeline that makes the separation between the gunshots and the Bellah crosstalk even greater, thus favoring their argument. Of course my concern has been with getting the facts right, not my opinions. My criticism was of their false claim of having a playback that was free of skips, and ultimately, their unscientific approach of providing calculated values without error terms. What all this means is that as a measure of recorded time for the interval from GO to YOU, the Audograph playback time, which compensates for skips, is probably more accurate than the playback from the phonograph which does not. The bottom line is that the whole process of playing back the disc on the phonograph was a waste of time, from the standpoint of measuring recorded time. Undoubtedly, the Audograph machine was designed to record and playback the way it does, with the turntable moving instead of the stylus arm, for the very purpose of avoiding the problem that floating stylus arms are prone to: getting stuck in one groove indefinitely. Because they failed to take into consideration how the machine worked, all of the NRC, and now the IBM’s measurements of time intervals on the Ch-2 recording are wrong, not just the one I cited in 2001.

 

 

 

A problem that does affect the synchronization issue is the dispatcher’s time notations.

The dispatcher’s time notations provide an ideal data set for testing alternative timelines because they are strongly dependent on the passage of real time, yet completely independent of any events such as the assassination or the position of the putative gunshots. The dispatcher’s time notations were not meant as precise time markers, but rather were appended to the dispatcher’s broadcasts as part of the radio protocol. Hence the frequency and spacing of the time notations was dependent on the frequency and spacing of the broadcasts. According to Bowles (p. 318) each dispatcher had a digital clock for making the time notations. A linear regression analysis of the time notations provides an objective mathematical method for testing the playback time against real time, because the time notations must fall within 60 sec bins, and because any long time periods between successive time notations must be offset by an equal number of short periods, which when summed will give an average of 60 sec if playback time and real time are the same. The linear regression analysis measures the deviation between the notated time intervals and the playback intervals, if any.

 

 

Table 2.- Least squares regression of time intervals among dispatcher’s time notations against expected time with and without corrections for the offsets (Δ) noted in Table 1 and as discussed in the text. All times in speed corrected playback in seconds. Data from O’Dell (2003). Correlation Coefficient (r) = 0.99 for all three regressions.

 

DISPATCHER             EXP.                OBSERVED      ALTERNATIVE     LINSKER ET AL

NOTATIONS                                         TIMELINE            TIMELINE         TIMELINE

 

  12:30                      0                        0                        0                      0              

 

  12:31                    60                       94                      94                    94            

 

  12:32                   120                     121                    121                  152          

 

  12:34                   240                     213                    213                  244         

 

  12:35a                 300                     269                    293                  324

 

  12:35b                 300                     300                    324                  355

 

  12:36a                 360                     330                    354                  385

 

  12:36b                 360                     362                    386                  417

 

 

 SLOPE                                              .91                     .99                  1.07

 

 

The results of the regression analysis are shown in Table 2. There is a relatively poor fit between the time notations and the playback time, a slope value of 0.91. But this is expected and predictable because we know that playback time is not real time because of

the data in Table 1. Note that there is a 24 sec offset in the time lapse between Bellah’s crosstalk and the simulcast by the Dispatcher at 12:35. If we add back the 24 sec to the interval between the two cross-talks and recalculate the slope, we now find very close agreement between the playback time and actual time as indicated by the dispatcher’s time notations. Importantly, there is an even larger time offset that occurs between 12:30 and 12:32, a total of 31 sec. If that time offset was caused by an imposition on Ch-1 then it would explain the apparent lack of synchrony between the gunshots and the assassination implied by the juxtaposition of the HOLD crosstalk. For that reason, and without a shred of evidence, the IBM authors assume that the offset was caused by recorder stoppage on Ch-2. But, adding the time to the appropriate interval on Ch-2 throws the regression line out of whack, with a slope of 1.07. What the analysis shows is that between these two alternative timelines, one of which is consistent with the acoustically identified gunshots being synchronous with the assassination, and the IBM timeline, that is not, the first has the best fit to the dispatcher’s time notations. I provided this information to the IBM authors in correspondence. In their article (p. 224) they criticize my reasoning stating,

 

“To favor one time calibration over another, the investigator must show that one regression analysis is better than the other to a statistically significant extent, and this was not done.”

 

That is exactly correct, and that is exactly what they failed to do. As the promulgators of the assertion that their timeline demonstrates a lack of synchrony between the putative gunshots and the assassination, a result that would be incongruent with the acoustical evidence, the burden of proof is on them to show that their timeline is superior to the alternatives, and that the difference is statistically significant. They can’t because it isn’t.

There is a caveat here. All of the Ch-2 time intervals in Table 2 are from the FBI phonograph playback, which as I have explained, are probably wrong.

 

In conclusion, the new report by IBM does not address the core acoustical evidence. Instead it cites the old NRC report stating that it contains a wealth of detail and should be read in conjunction with the new report. In point of fact, the original NRC report misrepresents the core acoustical evidence and does so in a highly prejudicial manner. The new report complains that I relied on the erroneous parts of the NRC report, which implies that there were parts of the NRC report that were not erroneous. Their proposed timeline of Ch-2 based on the Audograph playback was wrong because they didn’t understand how the audograph machine functioned. Their proposed timeline of the Ch-2 phonograph playback was based on their claim that this playback is free of skips, which we now know is not the case. Their proposed timeline of Ch-1 was wrong because they thought that the recorder speed problem was attributable to Bowles tape recorder when it was actually due to the use of two different Dictaphone machines. Their reliance on synchrony based on the Decker crosstalk ignored the evidence that all of the cross-talks are offset from one another demonstrating that synchronization based on any one crosstalk pair is unreliable, especially if the crosstalk is significantly separated from the event in question. The statistical analysis was fudged. Their account of the radio system function, specifically with regard to the origin of the crosstalk and other ambient noises, was wrong. Both the old NRC and new IBM reports do make contributions to our knowledge of the DPD recordings. But to derive value from those contributions one must separate the hard data from their contorted conclusions. The PCC analysis of the Fisher crosstalk in particular, is an example of new information that entirely conforms to our understanding of the acoustical evidence and corroborates that President Kennedy was killed by a gunshot from the grassy knoll.

    

References cited

 

Barger JE, Robinson SP, Schmidt EC & Wolf JJ. Analysis of recorded sounds relating to the assassination of President John F. Kennedy. US Congress, House of Representatives, House Select Committee on Assassinations Proceedings vol. 8. 1979.

Weiss, MR & Ashkenasy, A. An analysis of recorded sounds relating to the assassination of President John F. Kennedy. U.S. Congress, House of Representatives, House Select Committee on Assassinations Proceedings vol. 8. 1979.

Linsker R, Garwin RL, Chernoff H, Horowitz P and Ramsey NF. Synchronization of the acoustic evidence in the assassination of President Kennedy. Science & Justice 45: 207-226. 2005.

O’Dell M. The acoustic evidence in the Kennedy assassination. 2003. Posted at http://mcadams.posc.mu.edu/odell/.

Thomas DB. 2001. Echo correlation analysis and the acoustic evidence in the Kennedy assassination revisited. Science & Justice 41:21-32. Errata. Science & Justice 41:132. 2001.

Thompson, J. Six Seconds in Dallas, 1967, p. 248. FBI exhibit Q220.

[Warren] Report of the President’s Commission on the Assassination of President John F. Kennedy, U.S. Gov. Print. Off. Wash. D.C. 1964, pp. 111-117.

Bowles JC. The Kennedy assassination tapes: a rebuttal to the acoustical evidence theory. Pp. 313-410, In Savage G, JFK First Day Evidence. Shoppe Press, Monroe LA.  1993.

Ramsey NF, Alvarez LW, Chernoff H, Dicke RH, Elkind JI, Feggeler JC, Garwin RL, Horowitz P, Johnson A, Phinney RA, Rader C, and Sarles FW.  Report of the Committee on Ballistic Acoustics. National Research Council (USA) Prepared for Department of Justice, Washington D.C. Report No. PB83-218461. 1983.