Sabato, Sonalysts, & Sophistry
by Don Thomas
For his new book entitled, “The Kennedy Half Century” political scientist Larry Sabato commissioned an analysis of the acoustical evidence from the assassination. The conclusion of that analysis according to Sabato (p. 243) was:
“Our research demonstrates that the police officer with the open microphone was traveling at a high rate of speed at the time the slow-moving presidential motorcade progressed through the streets of down-town Dallas.”
In point of fact, the sounds identified as the gunshots by the HSCA’s acoustical experts occur at a segment of the police recording when the motorcycle is idling along. The disconnect between reality and the perceptions, or at least the opinions stated in this new study, is breathtaking. The formal analysis is not in the book but is rather posted on Sabato’s website. Dated October 15th, 2013 and entitled, “Analysis of the Dallas Police Department Dictabelt Recordings related to the Assassination of President John F. Kennedy,” the authors are Charles Olsen and Scott Martin of the firm Sonalysts of Waterford CT. The conclusion of the “voluminous” 14 page report is stated on the last page as follows:
“In summary, we can draw the following conclusions:
1. These data uniformly indicate that the motorcycle with the open microphone was not part of the motorcade.
2. Therefore it is unlikely that the motorcycle was in a position to record the sounds of gunfire.
3. Based on these observations we conclude that the Dictabelt Recording is not applicable to the identification of assassination gunfire.”
The report is long on technology but short on science, unless one includes political science. Actually the appropriate analogy is to Creation Science. As an evolutionary biologist I recognize the M.O. The problem is not in the data, it is with the interpretation. There is no discussion, let alone refutation, of the core acoustical evidence. By core acoustical evidence I mean the specific data cited by the HSCA acoustical experts as the evidence which compelled them to conclude that the assassination gunfire was captured by the police recording system. Rather the weight of their argument rests on what is at best ambiguous evidence concerning extraneous noises.
In contrast the acoustical evidence of gunfire on the Dictabelt is rock solid science. Like all good science it is falsifiable. If it can be shown that the sounds on the dictabelt identified as gunshots don’t match to test gunshots fired in Dealey Plaza then the gunfire hypothesis will have been falsified. The match was demonstrated by the HSCA’s experts (BBN), confirmed by a second lab (Computer Science Dept. at Queens College) and then reconfirmed in a study by a firm (Sensimetrics) commissioned by Court TV, in spite of the latter’s attempt to show the opposite. If it can be shown that the matching patterns are not ordered with respect to sequence, spacing and trajectory, just as would be expected of a microphone traveling with the motorcade, the gunfire hypothesis will have been falsified. It was this order in the matching data that compelled BBN to conclude that the assassination gunfire is on the dictabelt.
In the years following the HSCA report there has been ample corroboration of the acoustical evidence. The audio sequence of gunfire exactly fits the video sequence of wounding. Filmed evidence shows that a motorcycle with a sticky microphone was at the acoustically predicted locations. A broadcast common to both police channels synchronizes the acoustically identified gunshots with the moment of the assassination. The fact that the Sonalyst report fails to discuss or acknowledge this evidence strongly suggests that its objective was not to assess the evidence, but to discredit the evidence. In fact, Sonalysts data actually supports rather than conflicts with the HSCA acoustical analysis.
The key evidence is shown in their Fig. 14 reproduced here.
The Sonalyst data shows that after traveling along at a steady clip the motorcycle motor decreased speed sharply a few seconds before the 14 minute mark. The clock reference is the playback time on the dictabelt recording. A few seconds after the 14 min 30 sec mark the motor returns to its previous rate of speed. The Sonalyst report notes that the sharp drop in speed is coincident with the utterance “I’ll check it.” The report’s authors failed to realize, or at least, failed to discuss, the significance of this drop in speed. BBN noted that the motorcycle motor noise decreased by 70% just a few seconds before the first detected gunshot. BBN was unaware of the crosstalk so they made no reference to the utterance. The Sonalyst Fig. 14 does indicate the time of the last shot which was 8.3 sec after the first. As did Sonalyst, BBN also reported that the idle period lasted for about 40 sec. Importantly, because the microphone which recorded the test shot that matched to the first suspect sound was on Houston street about 20 ft from the intersection, BBN hypothesized that the motorcycle officer had let off the accelerator in anticipation of making the turn on to Elm Street. The following diagram illustrates the position of the motorcycle as predicted by the acoustical evidence. The five circles indicate the position of the microphones that recorded test shots that matched to the dictabelt patterns.
Of the 18 motorcycles in the motorcade the photographic evidence eliminates all but one from being at the acoustically required positions indicated by the circles. Officer H.B. McLain is seen at position (a) before the shooting and at position (c) after the shooting. At approximately 28 sec after the fatal shot McLain is seen at position (c) just passing the parked motorcycle of officer Bobby Hargis. Hargis had stopped his cycle and advanced on the grassy knoll. He returned to his cycle arriving about the same moment that he was passed by McLain. The photo (below) also shows officer J.W. Courson converging at this point. In a film by Elsie Dorman, Courson is seen arriving at the intersection of Elm and Houston at 6 sec after the fatal head shot. He was thus at least 100 ft behind McLain at that time. Because McLain was traveling at idle speed, taking 40 sec to travel 270 ft, Courson quickly closed the gap. When Courson caught up with McLain just after passing Hargis the pair sped up and left Dealey Plaza in tandem. Thus, contrary to the claim that the motorcycle motor speed was inconsistent with the motorcade movements, the movements of McLain’s motorcycle as seen in the films and as required of the acoustical evidence, is a close fit. It is part of the corroborative evidence. Of course, this is not news to those knowledgable of the evidence. What is new is the additional relevant data from Sonalysts analysis. On page 11 the report states, “Divergent data around 14 minutes, 30 seconds suggests that there may be multiple motorcycles heard in this time frame.” This of course is the moment when McLain, Hargis and Courson’s motorcycles converge.
Another non-sequitur in the Sonalyst report concerns the sounds of sirens near the 16 minute mark. One does not need a sonograph to hear that there are multiple sirens and opposing Doppler effects. Some are approaching, passing or are passed by the open microphone. Sonalysts judges this evidence to be inconsistent with a microphone on a motorcycle with the motorcade. Although it is not explicitly stated, the reasoning seems to be that if the open microphone was traveling with the motorcade it should be traveling at the same speed and direction as the sirens, and thus no Doppler effect. To first state which should be obvious, the acoustical evidence does not require that the motorcycle be with the motorcade at the time of the sirens, only that it was in Dealey plaza at the time of the shots. In fact, because of the previous evidence we know that McLain did not leave the plaza with the motorcade. Rather he lingered behind, leaving the plaza from 15 to 20 sec behind the President.
The simplistic scenario that the motorcade traveled as a unit to Parkland Hospital at speeds of 80 mph as stated by Sonalyst is certainly not the case. The motorcade disintegrated in the wake of the assassination. Some motorcycles remained with the President, but others stopped in Dealey Plaza to search for the gunmen. Others like Courson and McLain slowed then tried to catch up. Forgotten in all this is the advance guard; a phalanx of motorcycles that cleared the road ahead of the motorcade. These had left Dealey Plaza by the time of the shooting and were presumably somewhere en route to the Trade Mart when Jesse Curry ordered the escort to proceed to Parkland Hospital. Did these, or other officers at the Trade Mart, turn back in order to rejoin the motorcade? It is difficult to conceive of any scenario to explain multiple sirens traveling in different directions otherwise. Moreover, two minutes after the shooting when the sirens begin is about where the motorcade would have turned off of the Stemmon’s freeway to take Harry Hines Blvd to the hospital, just after passing the trade mart, and a likely place for a convergence to occur. The data may be inconsistent with the scenario posited by Sonalyst, it is not obvious why multiple sirens passing or passed by the microphone is inconsistent with McLain’s catching up with the motorcade.
But there is a further consideration. Throughout the report Sonalyst makes reference to “the open microphone,” as if all of the ambient sounds were captured by the one relevant motorcycle. That a succession of microphones were capturing and transmitting sounds need not be stated. Whether Sonalyst knew, or cared not to acknowledge, that multiple microphones were open at the same time is not clear. The most obvious instance occurred when the dispatcher made a broadcast saying, “Attention all emergency vehicles” and the motorcycle motor can be heard humming along in the background. The dispatcher was not riding on the motorcycle, and the motorcycle was not doing wheelies in the communications office. These were two different microphones capturing and transmitting sounds simultaneously. The police radio system acted like a party-line. As long as the capture ratio of competing signals was close to unity they would be simultaneously transmitted. In contrast, the motorcycle motor cannot be heard during the sirens. Sonalyst reasoned that it was “extinguished” by the volume of the sirens. Perhaps so. It is not inconceivable that one or more of the sirens were picked up by a different microphone or microphones. The point is that there are multiple scenarios that explain the siren sounds including but not limited to McLain’s radio. We just don’t know how far behind McLain was when the sirens start. This same party-line capacity could account for the other various bells, whistles, even some of the motorcycle motor noises, that are captured over the system. McLain’s was not the only motorcycle with a radio, nor the only one with a sticky microphone. The problem was indigenous to the Motorola model radios.
In a similar vein Sonalyst considered the origin of the crosstalks as Ch-2 transmissions that were “acoustically coupled,” that is picked up directly through the air by “The open microphone.” It is that narrow reasoning that led them to conclude that the “I’ll check it” broadcast was not a crosstalk. Because it is attended by a heterodyne it is clear that it was a Ch-1 transmission. In fact the most likely scenario is that all of the crosstalks were Ch-1 transmissions that originated as broadcasts over Ch-2. As noted by Captain James Bowles, there were some 200 police officers gathered at the Trade Mart to provide security. Crosstalk results when radio units are close enough together to pick up and rebroadcast their own transmissions. In this case the utterances were broadcast over a Ch-2 transmitter and captured and rebroadcast by a Ch-1 transmitter. The reason that the “I’ll check it” broadcast produced a heterodyne is that it’s higher energy caused an overload of the circuit producing the heterodyne. This particular utterance is loud and clear on Ch-1 compared to the other crosstalks which are barely audible. If the broadcasts were transmitted over one unit and captured over two microphones, or transmitted by two units and captured by the same microphone, the units closer together would produce the stronger signal and account for the heterodyne. Discrediting this instance of crosstalk has long been a target of the critics because it establishes exact synchrony between the acoustically identified gunshots and the assassination. The crosstalk was identified as such by Captain Bowles who prepared the official transcripts and who knew the voices of the officers involved. Moreover a voiceprint comparison of the respective broadcasts supported the identification. The Sonalyst report claims that when they filtered out the heterodyne noise the broadcast then sounded like the words “Five-Seven.” To assess the credibility of this claim, let us state their hypothesis in full. One must believe that electronic interference at the time of the broadcast of the words “five-seven” distorted the signal in such a way as to cause the utterance to sound like the words, “I’ll check it.” So much like the words “I’ll check it” is this utterance that it was recorded as such by the police officer who prepared the transcripts. And thus, the fact that this distorted utterance sounding like “I’ll check it” occurs a few seconds before the acoustically identified gunshots on Ch-1 while on Ch-2 the original broadcast of the words”I’ll check it” are spoken two seconds before the marker for the assassination making it appear that the two events are simultaneous is just a diabolical coincidence. And this happened at a time when many utterances on Ch-2 were being cross-talked on to Ch-1!
Because the study focused only on the motorcycle motor and other noises, something or someone, perhaps Sabato, compelled Sonalysts to go back months later to find a way to explain away the sounds identified as gunfire by the HSCA acoustical experts. In an appendix to their report Sonalyst provides a catalog of white noises that could account for the impulses which BBN identified as gunshot noises. In all of this they are correct. It is even plausible that a few of the impulses were part of an oscillatory pattern rather than discrete events as echoes of a muzzle blast should be. But this argument cannot account for the match between the test shots and the suspect sounds. One does not need to provide specific alternatives in testing the null hypothesis. One just assumes that other explanations are possible. In an effort to diminish the significance of the matches Sonalysts complained that the HSCA experts had put too much emphasis on the timing and not enough emphasis on the amplitude of the impulses. Had they included amplitude in the scoring the matches would not have been achieved. This argument rested on Sonalysts misunderstanding of the effect of automatic gain control which rendered amplitude of the impulses as unreliable. The AGC does not just suppress the amplitude of loud sounds as they argued on p. 15, but amplified the weak ones. The suppression of the AGC initiated by the muzzle blast has a duration of about 50 msec, so it would not have suppressed the impulses 200 msec later as Sonalysts supposed.
The approach by BBN for separating signal from noise was to use a threshold amplitude based on the motorcycle motor noise level. In their key analyses BBN and WA considered only those impulses that exceeded the threshold. Their comparison between test shots and the suspect sounds was based strictly on matching the time history of the impulses. In an effort to make it appear that the matches were likely to be false positives Sonalysts argued, “The BRSW (=BBN) makes no mention of testing the binary correlation detector prior to employing it on the dictabelt recording. Such measurements would have yielded performance data for the detector.” Sonalysts also described the detector as a filter. This argument was the height of sophistry.
In reading this passage many persons might be led to think that the detector was some sort of machine or device. It is a mathematical formula found in standard statistics text books. Its performance is measured mathematically by calculating the significance value, p = probability. Besides, the conclusions of the acoustical experts did not depend on the matching alone but on the order in the matching data which is not mentioned by Sonalysts. The order in the matching data was the test that eliminated any but the remotest probability of the matches being false positives.
If Professor Sabato thought he was paying for an objective assessment of the acoustical evidence, he should demand his money back.