The HSCA Acoustical Evidence: Proof of a Second Gunman in the JFK Assassination

[Michael T. Griffith]

Yet not one acoustic expert has expressed any support for Dr. Thomas’s most important claim. That there is only a 1-in-100,000 chance that correlation found by Weiss and Aschkenasy correlates with the impulse at 145.15 by random chance. Not even Dr. Barger.

And not one acoustical expert has disputed Dr. Thomas's calculation, even though it first appeared in his article in a peer-reviewed criminal science journal in 2001, and even though that article drew international attention. Furthermore, no critic of the acoustical evidence has yet refuted Dr. Thomas's calculation.

I have asked you several times to explain why you reject Dr. Thomas's calculation, but so far you have refused to do so.

I am also still waiting for you to explain the locational correlations between the dictabelt impulse patterns and five of the test-firing shots. If the dictabelt impulse patterns were the result of static, no one would expect them to match the five test shots in the correct locational order, given that the odds of doing so are 1 in 120.

The NRC panel was smart enough, and dishonest enough, not to try to explain this remarkable order in the data: they just ignored it.

[Joe Elliott]

And not one acoustical expert has disputed Dr. Thomas's calculation, even though it first appeared in his article in a peer-reviewed criminal science journal in 2001, and even though that article drew international attention. Furthermore, no critic of the acoustical evidence has yet refuted Dr. Thomas's calculation.

I have asked you several times to explain why you reject Dr. Thomas's calculation, but so far you have refused to do so.

Many people write all sorts of stuff that I don’t read. At a minimum, Dr. Thomas must first receive some support from acoustic experts for me to look into it. It makes so sense for Dr. Barger, Dr. Weiss and Mr. Aschkenasy to remain silent on this if there was any truth to Dr. Thomas’s 1-in-100,000 odds claim.

Also, Weiss and Aschkenasy have not provided enough information for me to even begin to calculate the odds. I know they used a computer programs to calculate the expected waveform:
•   From various positions within a few feet of microphone 3 ( 4 )
•   From various positions of the shooter within a few feet of the 1978 Grassy Knoll position
And see which combinations of microphone positions and shooter positions matches the 1963 145.15 impulse.
But how many microphone positions did they simulate? How many shooter positions did they simulate? If they simulated 300 positions each, that gives 90,000 unique pairs of positions that the computer could try to match up. It seems like one should find some pretty good correlations from sheer chance if they checked out that many permutations. But Weiss and Aschkenasy never provided us with this basic information. So, without it, I don’t see how anyone can come up with a good probability calculation.

Question 1:

How many permutations of hypothetical shooter and motorcycle positions did Weiss and Aschkenasy simulate when they ran their computer program, to generate waveforms to compare with impulse 145.15?

Question 2:

How does one calculate the odds of Weiss and Aschkenasy coming up with such a good correlation, without knowing the number of permutations of hypothetical shooter and motorcycle positions they simulated on the computer ?

I am also still waiting for you to explain the locational correlations between the dictabelt impulse patterns and five of the test-firing shots. If the dictabelt impulse patterns were the result of static, no one would expect them to match the five test shots in the correct locational order, given that the odds of doing so are 1 in 120.

The NRC panel was smart enough, and dishonest enough, not to try to explain this remarkable order in the data: they just ignored it.

Ridiculous. I have explained out this order can come about many times without involving any kind of coincidence. You think others won’t notice this?

Question 3:

If BBN only checked:

between microphones 2 ( 4 ) and 2 ( 6 ) for the first shot,
between microphones 2 ( 6 ) and 3 ( 5 ) for the first shot,
between microphones 2 ( 10 ) and 2 ( 12 ) for the first shot,
between microphones 3 ( 4 ) and 3 ( 8 ) for the first shot,
between microphones 3 ( 5 ) and 3 ( 8 ) for the first shot,

won’t there be a much better chance than 1-in-120 for the “motorcycle position” for the correlations they do find to come up in the proper order?


Are you going to continue to ignore 3 simple questions and hope the other readers won’t notice?

[Michael T. Griffith]

I have mentioned the often-overlooked fact that the NRC panel concluded that there was only a 22% probability that chance caused the correlations between the dictabelt grassy knoll shot (the 145.15 impulse pattern) and the test-firing grassy knoll shot. Yes, a 22% probability of chance is higher than a “5% or less” probability of chance, but it is still not a very good chance.

If someone offered you an investment that had only a 22% chance of earning a profit, you would decline it. If a candidate received only 22% of the vote in an election, he would be judged to have lost in an enormous landslide. If a mechanic gave you an estimate of $4,000 for a car repair and added that there was only a 22% chance the repair would actually fix the problem, you would say “no thanks” (unless you were very rich). If you discovered that you had cancer and that you had only a 22% chance of beating it, you would start getting your affairs in order in anticipation of your probable death. Any way you slice it, a 22% chance is not a good bet.

Of course, a 22% probability that chance (random noise) caused the correlations equals a 78% probability that gunfire caused the correlations. The fact that the NRC panel, even after making several dubious assumptions in their calculations, said there was a 78% chance that the 145.15 impulse pattern was caused by gunfire gives us some idea of just how strong and solid the WA sonar analysis was. 

I quote from the NRC panel’s report (keep in mind that “P” here equals the probability of chance, and that P=0.223 means a probability of chance of 22%):

Thus 10 coincidences among 12 predicted echoes and 14 impulses out of 45 windows should be adjusted to 8 coincidences among 11 predicted echoes and 12 impulses out of 90 windows. If we now reduce the number of coincidences by 7 to make a conservative adjustment for the “free” parameters, we would have 1 coincidence among 4 predicted echoes and 5 impulses out of 83 windows. The hypergeometric probability calculation would then yield a significance level of P=0.223, which is not at all impressive in contrast to the claim that P=0.053. However, this adjustment may be unduly conservative. (NRC panel report, p. 40)

Did you catch the last sentence: the NRC panel’s “adjustment” may have been “unduly conservative”? This is an understatement. Dr. Thomas explains just how badly and errantly the NRC panel played with the numbers in order to get the probability of chance up to 22%:

But, because of using the incorrect values for M and i, BBN had grossly overestimated the probability that a random burst of noise could have matched the test pattern. This was obviously an honest mistake because the error worked against their own conclusion by increasing the probability of coincidence by chance. It is interesting that the NRC panel caught this error and knew that the correct number of time slots was 90 and not 45. It proves that they did understand that the 190 millisecond gap between the two 90 millisecond bursts occurred in the middle of the impulse pattern and were not two completely separate patterns as they had asserted in their critique of the HSCA analysis. . . .

Aware of the error made by BBN in assigning the value of M, the NRC panel looked for errors in the values of N, n and i. In scrutinizing the echo path data they noticed a discrepancy in the echo matching procedure. Among the 12 pertinent echoes in the Dealey Plaza test shot there were some that arrive at the microphone quite close together. One pair of echoes, designated 19 and 20, were separated by only 3 milliseconds on the test pattern. Applying the sonar model to the echo delay times predicted that these two echoes would have arrived precisely coincident with one another when the microphone was moved analytically to the predicted motorcycle position. In fact, there is only one spike in the evidence pattern at the corresponding echo delay location. Weiss and Aschkenasy scored the two echoes as separate coincidences giving a contribution of 2, 2, and 2 to the values of N, n and I respectively.

The NRC panel took the position that because there really is only spike in the evidence pattern in this position, and one in the predicted pattern, the contribution to N, n and i should have been 1, 1, and 1 respectively. Because there really are two echoes involved, the assigned values were not really an error on the part of Weiss and Aschkenasy. Rather, of the two reasonable alternative scoring procedures the NRC procedure was simply the more conservative. Adopting the conservative approach gives a number set of {90, 13, 11, 9}.

A similar problem involved echo numbers 23 and 24. On the test tape the impulses are separated by only 5 milliseconds. The sonar model predicts their arrival will be much closer, though not exactly coincident, approximately 1 millisecond apart, when the microphone is moved to the predicted motorcycle position. Likewise, the two echoes on the evidence tape at this echo delay location are about 1 millisecond apart. Again, Weiss and Aschkenasy scored the two echoes as contributing values of 2, 2, and 2 to the parameters N, n and i, respectively.

But the NRC panel argued that the coincident windows should be non-overlapping, and because the windows are 2 milliseconds wide the scoring should count only one coincidence not two. Inasmuch as there really are two echoes on the test tape, two separate spikes on the evidence tape, and two predicted echoes on the sonar model, the reduction from two to one coincidence in this case seems overly conservative.

In fact, the NRC panel scored the contributions to the parameters N, n and i as I, 2 and I, respectively, in effect, insisting on counting an echo for which they would not allow a match. Even accepting the overly conservative approach to scoring which insisted on nonoverlapping windows, their assigned score was clearly inappropriate. By the NRC scoring the number set was now {90, 12, II, 8}.

In this manner the NRC panel whittled away at the data set. Most of this whittling was accomplished by further reducing the number of allowed coincident impulses.

The NRC panel identified seven assumptions made by Weiss and Aschkenasy in constructing their sonar model. These were: that the time of the muzzle blast was coincident with the first impulse on the evidence tape, that the air temperature was 65 degrees F, that the tape recorder speed was slow by 4.3 percent, that the location of the motorcycle when the muzzle blast arrived was 72 ft. west of the intersection, that the motorcycle was traveling at 11 mph, and that the shooter was 8 ft. west of the corner of the fence on the grassy knoll. The final calculation of the odds depends on how one chooses to account for these assumptions. Were these reasonable assumptions, or had Weiss and Aschkenasy just picked out numbers that would give the best fit?

There is a formal way to do this. When a parameter is unknown, its value is said to be a "free variable." For example, the HSCA analysts recognized that the motorcycle position was a free variable and had thus adjusted their calculation of p by multiplying by a factor of 180 for the 180 possible positions around microphone 4(3). Alternatively, one can adjust for free variables by deducting directly from the values of the parameters in the formula. In statistical analysis these corrections of a parametric value are called "degrees of freedom."

The NRC panel decided that the assumptions in the sonar model involved 7 degrees of freedom and therefore deducted 7 coincident pairs from the data set. The number set which the NRC panel wound up with by deducting 7 for each parameter, beginning with their "corrected" number set of {90, 12, 11, 8} was {83, 5, 4, 1}. One assumes that they left the lone coincident because if they had whittled away at the data any further they would have left themselves without any data to argue from. The probability given by this number set calculates out to 20.7 percent (the NRC reported 22 percent). The estimate of 22 percent was much less impressive than the industry standard of 5 percent. The panel had attained its goal - final score: NRC 22, HSCA 5.

But, again, the validity of each of the assumptions, and the manner in which the adjustments were applied to the number set, is questionable. Certainly, the motorcycle position was a free variable. Weiss and Aschkenasy had, in essence, "selected" the position which gave the best match between the test and evidence patterns. Because the motorcycle position was free to vary in two dimensions, the NRC panel rightly deducted two degrees of freedom.

The shooter position was also an uncertain parameter and therefore a free variable. The NRC panel also figured this assumption was worth two degrees of freedom. But, in reality, the shooter's position on the grassy knoll was not free to move in two dimensions, but only one. The shooter could not have been any distance away from the fence, and thus, was only free to move in one dimension, i.e., along the fence.

But the other parameters cited by the NRC were not free variables at all. In fact, had these factors not been included, the analysis could be fairly criticized. The velocity of the motorcycle, the air temperature, and the tape recorder speed, all factors in the sonar model, were actual measurements, not unknowns.

Weiss and Aschkenasy had used values for these parameters such as the air temperature and motorcade speed, provided by BBN and the HSCA staff, all of which were derived from evidence independent of the acoustical analysis. They did find that a small adjustment of tape speed of 4.3 percent, rather than BBN's measurement of 5%, gave the best fit. Nevertheless, it seems unduly conservative to insist that these adjustments be treated as free variables as if they were numbers plucked out of the air to obtain a fit, rather than measured parameters. Even allowing for the validity of all of the NRC panel's adjustments, one was duplicated.

The NRC panel eliminated the erroneous adjustment for the M value by BBN, but failed to recognize that Weiss & Aschkenasy had already made the adjustment for the alignment of the muzzle blast with the first impulse. Just putting back one for the overt error in the deduction for this alignment gives a data set of {84, 6, 5, 2}. The probability of this coincidence set is 3.7 x 10^-2 or, about 3.7 percent. Even better than Weiss and Aschkenasy's original five percent!

The most appropriate estimate of the true probability would assign two degrees of freedom for the motorcycle location, one for shooter location, one for the alignment of the muzzle blast, and eliminate one pair of echoes and their coincidence (from pairs 19 and 20), giving a final number set of {86, 10, 8, 6} which calculates to 1.12 x 10^-5, or, 100,000 to 1, against. (Hear No Evil, pp. 629-632)

[Joe Elliott]

I have mentioned the often-overlooked fact that the NRC panel concluded that there was only a 22% probability that chance caused the correlations between the dictabelt grassy knoll shot (the 145.15 impulse pattern) and the test-firing grassy knoll shot. Yes, a 22% probability of chance is higher than a “5% or less” probability of chance, but it is still not a very good chance.

If someone offered you an investment that had only a 22% chance of earning a profit, you would decline it. If a candidate received only 22% of the vote in an election, he would be judged to have lost in an enormous landslide. If a mechanic gave you an estimate of $4,000 for a car repair and added that there was only a 22% chance the repair would actually fix the problem, you would say “no thanks” (unless you were very rich). If you discovered that you had cancer and that you had only a 22% chance of beating it, you would start getting your affairs in order in anticipation of your probable death. Any way you slice it, a 22% chance is not a good bet.

Of course, a 22% probability that chance (random noise) caused the correlations equals a 78% probability that gunfire caused the correlations. The fact that the NRC panel, even after making several dubious assumptions in their calculations, said there was a 78% chance that the 145.15 impulse pattern was caused by gunfire gives us some idea of just how strong and solid the WA sonar analysis was. 

I quote from the NRC panel’s report (keep in mind that “P” here equals the probability of chance, and that P=0.223 means a probability of chance of 22%):

Did you catch the last sentence: the NRC panel’s “adjustment” may have been “unduly conservative”? This is an understatement. Dr. Thomas explains just how badly and errantly the NRC panel played with the numbers in order to get the probability of chance up to 22%:

The only type of acoustic test worth evaluating is one that, at a minimum, checks the correlation of:

•   every 1963 impulses, all 7 of them,
with:
•   every 1978 test shot, all 12 of them,
•   every 1978 microphone position, all 36 of them,

and gives a complete list of all found correlations that exceed a certain threshold.


It is difficult to calculate the probability that a found correlation is the result of chance. One expert might say it is 22%, another 5%, another 0.001%. Who do you belief?


But if strongest correlations that don’t conflict with each other and the complete list of the strongest correlations look something like:

*********** Warning, False Data just to make a point ***********
shot 1 fired from TSBD, fired at 200 feet from Houston Street, recorded at 100 feet from Houston Street,
shot 2 fired from TSBD, fired at 220 feet from Houston Street, recorded at 120 feet from Houston Street,
shot 3 fired from TSBD, fired at 240 feet from Houston Street, recorded at 140 feet from Houston Street,
shot 4 fired from knoll, fired at 300 feet from Houston Street, recorded at 200 feet from Houston Street,
shot 5 fired from TSBD, fired at 320 feet from Houston Street, recorded at 220 feet from Houston Street,

this would be consistent with the recording being made at Dealey Plaza, since:
•   none of the correlations contradict one another, like finding correlations for the same shot from both the TSBD and the knoll,
•   none of the correlations contradict a likely fact, like targets that do not approximately correspond to the location of the limousine,
•   none of the correlations contradict a likely fact, like motorcycle locations that don’t match up with a motorcycle proceeding forward at all times

There must be reasonable correlations for all three, the locations of the rifle, the location of the target, the location of the motorcycle. Getting only two of three or one of three correct is no good.


However, if we get strong correlations that contradict each other or highly probable facts, like:

•   correlation of the same 1963 impulse with both a test shot from the TSBD and the knoll,
•   correlation with the target location not matching the approximate limousine location,
•   correlation of the motorcycle position indicating a reversal of direction,

then we know we have found no good evidence that the recording was made at Dealey Plaza. This is certainly not the case of the BBN correlations found in August 1978 and recorded in BBN Exhibit F-367.

It doesn’t matter if some expert claims the odds of getting a correlation is only 5%, or 1%, or 0.001%. We know these correlations are invalid because they contradict one another.


The Weiss and Aschkenasy study have no such problems, but it was impossible for it to have these problems. Because they didn’t run correlations of different 1963 impulses, with different 1978 test shots, recorded at different positions. They only compared one 1963 impulse with one 1978 test shot recorded from one 1978 microphone position. Guarantying no correlation contradictions would be found, regardless of whether the recording was made at Dealey Plaza, or the Trade Mart Center, or anywhere else. Making the Weiss and Aschkenasy tests a waste of time.


One final point, I like the BBN study a lot better than the Weiss and Aschkenasy study. If someone makes the claim that “The odds of any of the BBN correlations being a result of random chance are 1-in-100,000”, it is easy for me to refute this. Because many of the correlations they found contradict each other. Of course, if amazingly enough, none of the correlations contradicted each other, this claim would have to be taken seriously as at least being possibly true.

But if one claims “The odds of the Weiss and Aschkenasy correlation being a result of random change are 1-in-100,000”, it is not easy to automatically refute this. Because the Weiss and Aschkenasy made such limited comparisons that it was impossible for any contradiction to be found, since they were only comparing one 1963 impulse, with one 1978 firing test, fired from one position, fired at one target. Instead, I have to use other means of discounting this claim. Like pointing out that no acoustic expert has expressed any support for this claim. True, no acoustic expert has expressed any opposition to this claim either. But scientists don’t usually comment on the claims of cranks talking outside their own field of expertise.

[Michael T. Griffith]

Lately I have been re-reading the NRC panel’s report. I began to do so mainly to double-check the exact wording they used when they said there was only a 22% probability that chance (random noise) caused the correlations between the dictabelt grassy knoll shot and the test-firing grassy knoll shot.

As I documented in my previous post, the NRC panel used dubious, invalid values in order to calculate the probability of chance for the grassy knoll shot as being 22%. When I first read this way back when, I thought, “But that means you’re admitting there’s a 78% probability that the 145.15 impulse pattern was caused by gunfire from the grassy knoll!” The NRC panel could not even bring themselves to state this.

Anyway, in re-reading the NRC panel’s report, I can see why the HSCA acoustical scientists treated it so dismissively. The report does not even mention, much less try to explain, the locational order in the matches between the dictabelt gunshots and the test-firing gunshots. The report does not even mention the fact that the dictabelt gunshot impulse patterns indicate the microphone was moving at the same average speed as the motorcade. The report does not even mention, much less explain, the fact that N-waves not only appear in the dictabelt gunshots when they should, and do not appear when they should not, but that when they appear, they do so in the correct order and interval in relation to the muzzle blast. I mean, the report literally says nothing—not one word—about any of these correlations.

What makes the report’s silence even more discrediting and damning is fact that the report indicates that the NRC panel understood that such correlations would be expected to be found in the dictabelt impulse patterns if assassination gunfire caused those patterns. For example, the NRC panel’s report admits that the Dealey Plaza test firing proved that for each shot fired in Dealey Plaza there will be “a characteristic and complex pattern of echoes”:

For each combination of shooter location and microphone location, there is a characteristic and complex pattern of echoes in the recording of the test shot. . . . (NRC panel report, p. 8 )

Yet, the panel chose to say nothing about the fact that the HSCA acoustical scientists documented that at least four of the dictabelt impulse patterns contain the “characteristic and complex pattern of echoes” of four of the Dealey Plaza test-firing shots.

Revealingly, the report shows that the NRC panel understood that the shock wave (N-wave) comes before the muzzle blast and that the muzzle blast is followed by gunfire echoes. The report mentions this fact in the same paragraph quoted above:

For each combination of shooter location and microphone location, there is a characteristic and complex pattern of echoes in the recording of the test shot, because after the first sounds travel by a direct path to the microphone (usually the bullet shock wave and the muzzle blast sound), subsequent sounds arrive (due to echoes from buildings and other large objects) with varying delays, depending on the length of the path they have taken. (NRC panel report, p. 8 )

Well, well, so the panel members did know that there is a defined order in how and when the sounds of the shock wave (N-wave), the muzzle blast, and the N-wave and muzzle-blasts echoes (“subsequent sounds”) will travel and be recorded—as long as the microphone is in position to record them, of course. The HSCA acoustical experts spent a fair amount of time talking about this fact. They explained—and one can confirm this in ballistic-acoustics literature—that the N-wave will always come before the muzzle blast and will do so in a predictable interval, that the muzzle blast will come behind the N-wave, and that N-wave and muzzle-blast echoes will come last.

Yet, the panel chose to say nothing about whether or not N-waves appear on the dictabelt recording. Thus, needless to say, the panel said nothing about the fact that N-wave impulses occur in the dictabelt gunshots when they should, and do not occur when they should not; that they occur in the correct order and interval in relation to the muzzle-blast impulses; and that echoes of them occur afterward.

Another line of evidence that the NRC panel’s report ignores is the windshield-distortion correlations. The failure to even mention, much less address, the windshield-distortion correlations is suspicious and revealing. The HSCA acoustical scientists conducted a test to determine the acoustical characteristics of windshield distortion, because they correctly suspected that such distortion would occur whenever the windshield was between the shooter and the microphone. And, lo and behold, they found that windshield distortion occurs in every dictabelt gunshot where it should and does not occur in the one dictabelt gunshot where it should not.

The NRC panel was surely aware of the windshield-distortion correlations. The HSCA acoustical experts talked about them during the hearings, and the HSCA report highlighted them:

Weiss and Aschkenasy also considered the distortion that a windshield might cause to the sound impulses received by a motorcycle microphone. They reasoned that the noise from the initial muzzle blast of a shot would be somewhat muted on the tape if it traveled through the windshield to the microphone. Test firings conducted under the auspices of the New York City Police Department confirmed this hypothesis. Further, an examination of the dispatch tape reflected similar distortions on shots one, two, and three, when the indicated positions of the motorcycle would have placed the windshield between the shooter and the microphone. On shot four, Weiss and Aschkenasy found no such distortion. The analysts' ability to predict the effect of the windshield on the impulses found on the dispatch tape, and having their predictions confirmed by the tape, indicated further that the microphone was mounted on a motorcycle in Dealey Plaza and that it had transmitted the sounds of the shots fired during the assassination. (HSCA report, pp. 74-75)

Perhaps the NRC panelists chose not to even mention these remarkable correlations, as well as the other correlations, because they feared that their all-or-nothing reliance on the Decker “hold everything” crosstalk would look lame by comparison. Perhaps the panelists feared that many people would logically observe, “Since you are claiming that the dictabelt gunshot impulse patterns are not gunshots and were not even recorded during the assassination, you are asking us to believe that the correlations documented by the HSCA acoustical experts are all just astounding coincidences. That seems very hard to believe.”

We can understand why the NRC panel’s report is so bad when we realize that the panel’s leading member, its driving force, was Dr. Luis Alvarez, who was an ardent Warren Commission defender and who was caught fabricating his own test data to defend the commission’s claims. When BBN acoustical scientist Dr. James Barger, one of the HSCA acoustical experts, met with the NRC panel, Alvarez told him that he was going to vote against the HSCA findings no matter what Dr. Barger said. Dr. Donald Thomas:

The NRC report is replete with false and erroneous statements. James Barger, the lead scientist with BBN, met with the panel in an effort to explain his analysis and defend his conclusions. But, according to Barger, Alvarez told him it didn't really matter what he had to say, he (Alvarez) was going to vote against the HSCA findings anyway. (Thomas, Hear No Evil, p. 619)

[Joe Elliott]

Lately I have been re-reading the NRC panel’s report. I began to do so mainly to double-check the exact wording they used when they said there was only a 22% probability that chance (random noise) caused the correlations between the dictabelt grassy knoll shot and the test-firing grassy knoll shot.

As I documented in my previous post, the NRC panel used dubious, invalid values in order to calculate the probability of chance for the grassy knoll shot as being 22%. When I first read this way back when, I thought, “But that means you’re admitting there’s a 78% probability that the 145.15 impulse pattern was caused by gunfire from the grassy knoll!” The NRC panel could not even bring themselves to state this.

This 78% probability is true. If we cherry pick what facts to consider. But when we consider all factors, the probability drops:
•   Several people are within 50 feet of the fairly exposed grassy knoll sniper position, Mr. Zapruder, Ms. Sitzman, the young black couple who dropped a coke bottle, others on the steps below them. A strange shooting position where one would think a shooter wouldn’t want to be on ground where anyone could walk up to them.
•   The Grassy Knoll was filmed during the shooting but no clear image of a life size human who is a possible shooter is seen.
•   The difficulty a shooter would be giving themselves by choosing a position where they would have to fire a shot at a moving target that is moving at almost at a right angle relative to them, giving the target a near maximum angular velocity.
•   No shells recovered from the grassy knoll.

Anyway, in re-reading the NRC panel’s report, I can see why the HSCA acoustical scientists treated it so dismissively. The report does not even mention, much less try to explain, the locational order in the matches between the dictabelt gunshots and the test-firing gunshots. The report does not even mention the fact that the dictabelt gunshot impulse patterns indicate the microphone was moving at the same average speed as the motorcade. The report does not even mention, much less explain, the fact that N-waves not only appear in the dictabelt gunshots when they should, and do not appear when they should not, but that when they appear, they do so in the correct order and interval in relation to the muzzle blast. I mean, the report literally says nothing—not one word—about any of these correlations.

Yes, but I have provided an explanation for this many times, which you ignore. If they checked the eastern microphones for the early shots, the western microphones locations for the later shots, the probability drops greatly down from 1-in24 (4 shots) or 1-in-120 (5 shots). The correlations found should match the scenario of a motorcycle steadily advancing along Houston and Elm. The arguments for this are:
•   The difficulty of completing all 3,024 correlation checks in 10 days, hence the need to limiting the checks to the most probable microphone locations the motorcycle most likely would be in at that time.
•   The lack on any statement, by Dr. Barger, or in the BBN report, where they argue that the probability of the four shots being in the proper order would be only 1-in-24. They would, of course, omit this obvious argument, if they knew there was an alternative explanation for this apparent order.

[Michael T. Griffith]

I have already noted the dubious assumptions that the NRC panel had to make to get their feeble 22% probability that random noise caused the dictabelt grassy knoll gunshot, as opposed to the “5 percent or less” probability calculated by Weiss and Aschkenasy (WA).

But it is also important to note that the NRC panel ignored—simply flat-out ignored—a key qualification that WA stipulated regarding their “5% or less” probability of chance: WA stipulated that this calculation was “highly conservative” and that the actual probability of chance was “considerably less” than 5%.

WA explained that their “5% or less” calculation assumed that impulses could only occur in the two intervals in which echoes were observed to occur in the test-firing grassy knoll shot. These two intervals were from 0 to 85 milliseconds and from 275 to 370 milliseconds.

The grassy knoll test shot produced a distinctive pattern in which the echoes arrived in two clusters. The first cluster arrived in the first 85 milliseconds and consisted of echoes from structures facing Elm Street. The first cluster was followed by a gap of 190 milliseconds, corresponding to the open space at the intersection. The second echo cluster arrived in the last 95 milliseconds (275 to 370) and originated with the structures on Houston Street. This is why WA assumed in their analysis that impulses could only occur during the timespan of these two timeframes, i.e., 180 milliseconds.

But, if the dictabelt impulses were not caused by gunfire, then the timespan during which impulses could have occurred more than doubles: it goes from 180 milliseconds to 370 milliseconds. Why? Because there is a 190-millisecond interval between the two intervals of 0-85 and 275-370 milliseconds. Obviously, if you more than double the timespan for impulses to occur, this vastly reduces the probability that random noise caused the dictabelt grassy knoll shot.

Here is how WA explained this in their report:

The high degree of correlation between the impulse [the dictabelt grassy knoll shot] and echo sequences [of the grassy knoll test shot] does not preclude the possibility that the impulses were not the sounds of a gunshot. It is conceivable that a sequence of impulse sounds, derived from non-gunshot sources, was generated with time spacings that, by chance, corresponded within one one-thousandth of a second to those of echoes of a gunshot fired from the grassy knoll. However, the probability of such a chance occurrence is about 5 percent. This calculation represents a highly conservative point of view, since it assumes that impulses can occur only in the two intervals in which echoes were observed to occur, these being the echo-delay range from 0 to 85 milliseconds and the range from 275 to 370 milliseconds. However, if the impulses in the DPD recording were not the echoes of a gunshot, they could also have occurred in the 190-millisecond timespan that separated these two intervals. Taking this timespan into account, the probability becomes considerably less than 5 percent that the match between the recorded impulses and the predicted echoes occurred by chance. (8 HSCA 32)

[Michael T. Griffith]

This 78% probability is true. . . . [SNIP]

No, it is not true. Dr. Thomas has explained in great detail why it is false. I quoted Dr. Thomas's entire rebuttal to the NRC panel's 78% figure, but you snipped and ignored the rebuttal, and then treated us to more of your diversionary observations and errant speculation.

You pull this stunt all the time: You ignore arguments that you can't explain; you type several paragraphs of irrelevant observations and flawed speculation; and then you pretend that you have dealt with the arguments.

Quote from: Michael T. Griffith on October 13, 2020, 02:50:07 PM
Anyway, in re-reading the NRC panel’s report, I can see why the HSCA acoustical scientists treated it so dismissively. The report does not even mention, much less try to explain, the locational order in the matches between the dictabelt gunshots and the test-firing gunshots. The report does not even mention the fact that the dictabelt gunshot impulse patterns indicate the microphone was moving at the same average speed as the motorcade. The report does not even mention, much less explain, the fact that N-waves not only appear in the dictabelt gunshots when they should, and do not appear when they should not, but that when they appear, they do so in the correct order and interval in relation to the muzzle blast. I mean, the report literally says nothing—not one word—about any of these correlations.

Yes, but I have provided an explanation for this many times, which you ignore. . . . [SNIP]

One, your explanation is patently absurd. Two, I have not ignored your explanation but have responded to it several times, noting that, among other things, it simply ignores the details of the sonar analysis and the fact that the sonar analysis was able to simulate closer microphones and 180 positions.

"But, but, but . . . they didn't check this microphone or that microphone," etc., etc. Why don't you deal with what they did do in the sonar analysis? Why don't you deal with Dr. Thomas's point-by-point refutation of the NRC panel's bogus value assumptions for their 78% probability of chance?

•   The lack on any statement, by Dr. Barger, or in the BBN report, where they argue that the probability of the four shots being in the proper order would be only 1-in-24. They would, of course, omit this obvious argument, if they knew there was an alternative explanation for this apparent order.

What??? You still have not read the BBN report, have you?

In point of fact, the BBN scientists were tremendously impressed with the locational correlations between the dictabelt gunshots and the test-firing gunshots. They determined that the probability that chance caused those correlations was “less than 1%.” Figure 22 in the BBN report shows the microphone positions along the motorcycle route where high correlations were obtained. The BBN scientists referred to this figure in explaining why there was less than a 1% probability that chance caused the time-distance correlations. I quote from the BBN report, which you really should read some day:

Even a brief glance at Fig. 22 shows that the microphone locations that correspond to correlations at the three times after the first impulse tend to progress uniformly forward along the motorcade route. This conclusion can be quantified statistically by the chi-square test. If the motorcycle were not moving through Dealey Plaza at the time of the assassination, the distance along the motorcade route would be a meaningless coordinate, and the microphone locations for the correlations that exceed the detection threshold would occur at random. When the chart in Fig. 22 is partitioned into a 2 x 2 table by separating time at 5 sec and distance at 250 ft, we find 1, 6, 8, and 0 correlations in the four sections reading from left to right, top to bottom. But the expected number of correlations to be found in these four sections, if the correlations occurred at random, are 4.2, 2.8, 4.8, 3.2. The value of chi-square for the observed and expected values is equal to 11.4. There is only 1 degree of freedom in this 2 x 2 table, and the probability that this large value of chi-square could occur at random is less than 1%. Therefore, there is little doubt that the distance coordinate is meaningful, and we conclude that the motorcycle was moving through Dealey Plaza and did, in fact, detect the sounds of gunfire. (BBN report, 8 HSCA 104)

And what did the NRC panel have to say about this powerful evidence? They argued that the BBN scientists had erred, that the BBN value of P<0.01 (i.e., less than 1%) should actually be P=0.07 (i.e., 7%), and that therefore the “significance of the layout” indicated by Figure 22 is “considerably reduced” (NRC report, p. 37). Alright, so instead of the probability of chance being less than 1%, the NRC panel said it is 7%.

Now, many readers will say, “Well, okay, but that means the NRC guys admitted that the probability that chance caused those correlations is only 7%, which means the probability that the police tape was recorded by a motorcycle in Dealey Plaza is 93%.” Indeed.

The NRC panel made no effort to explain the significance of the fact that their own calculation found a 93% probability that the locational correlations occurred because the impulse patterns on the police tape were recorded by a motorcycle in Dealey Plaza. In fact, they did not even specifically mention this. They simply noted that they determined the probability of chance was 7% and acted as though they had dealt a strong blow to the BBN report. Granted, a 7% probability of chance is much more than a <1% probability, but it is still an extremely low probability.

On a side note, the BBN report explains that not every recorded shot would have an N-wave (shock wave) in its impulse patterns because the microphone that recorded the shot was not in position to record it; but, if the microphone were in position to record the N-wave, the N-wave would be “a significant part” of the echo pattern:

In Sec. 2, we described how the shock waves generated by rifle bullets would be sufficiently loud at some microphone positions to become a significant part of an echo pattern. The speed of the bullet is important, because it determines the difference in time between perception of the shock waves and perception of the muzzle-blast waves. (BBN report, 8 HSCA 57)

As noted in previous posts, this is exactly what we see in the acoustical evidence: An N-wave appears in those dictabelt shots that were recorded when the motorcycle was in position to record the N-wave, and no N-wave appears in those shots where the motorcycle was not in position to record it (BBN report, 8 HSCA 49-50).