National Academy of Sciences (NAS)

Report of the Committee on
Ballistic Acoustics






Washington, DC 1982




This Appendix discusses the results of a study of the times of occurrence of certain key events on Channels I and II. The times were determined in two ways: 1) by listening to the Bowles tapes and by measuring the strip charts (in Figures C-1 and C-2) of the tape recording signal levels as a function of time; and 2) by listening to tape recordings made from the original Gray Audograph and Dictaphone records.


C-1. Analysis of Bowles Tapes

The Channel II transmission "you want me . . . Stemmons"* that occurs about 200 seconds after the transmission by Chief Curry, "Go to the hospital," is clearly audible and intelligible on Channel I. It provides a common reference point for synchronizing the Channel I and II tapes, and we can use it to determine whether the events on Channel I identified by BRSW/WA as shots occurred before or after Chief Curry broadcast his instructions, "Go to the hospital." If these events occurred after Chief Curry's instructions, they could not be the assassination shots.

[*A transcript of the relevant portions of both tapes appears at the end of this appendix (Tables C-2 and C-3). It was obtained from J. C. Bowles. The time used by Bowles is retained on the transcript even though it differs from the one favored by the Committee.]

The transmission "hold everything . . ." on Channel I coincides in time with the last of the events BRSW/WA identified as shots. The strip chart timings provide evidence to support an explanation of how this transmission could have occurred at the same point in real time as the matching "hold everything . . ." transmission on Channel II. They depend only on whether or not the instructions, "Go to the hospital," preceded the events identified by BRSW/WA as shots.


Channel I Recording

Measurement of the interval on Channel I between "you want me . . . Stemmons" and the conjectured shots is straightforward. The logging recorder (Dictaphone) ran continuously over the time of interest (even though it was sound-actuated), and the tape recording that we used for our measurements shows no evidence of skips, repeats, gaps, halts or similar artifacts that would affect the timing. Table C-1 gives the times of the transmissions of interest to us on Channel I.


Channel II Recording

Measurement of the time intervals on Channel I between "Do you want me . . . Stemmons" and "Go to the hospital" is more difficult. The logging recorder (Gray Audograph) was sound-actuated and did not operate continuously. So the actual record is shorter than real time. There are repeats in the Bowles tape recording that occurred when the Gray Audograph playback stylus jumped back to a previous groove in the record much as the stylus on a scratched phonograph record often does. The tape recordings made available to us initially were in two segments, with a break occurring between the two transmissions of interest. The first segment was recorded at a speed different from that of the second segment. All of these artifacts required compensation in order to obtain an accurate determination of the interval between the two transmissions. Compensation was done as follows:


a. Gap

Barger had access to an unbroken recording of the entire interval of interest from which he was able to show that a section of the original Channel II recording 0.4 seconds long had not been captured on the recordings we initially used.


b. Speed Compensation

The relative speed of the two segments of tape can be estimated from the sound spectrographs discussed in Appendix B. From this analysis we determined that times measured from the first segment of Channel II, which we designate Channel IIB, had to be multiplied by a factor of 1.06 to convert them to the time units of Channel IIB, the second tape recorder segment. Further, Barger and Weiss ascertained by an analysis of tones on Channel I and Channel IIB that these two tape recordings were made at essentially the same speed and that no correction was necessary to convert Channel IIB times to equivalent Channel I times.

BRSW reported that the Channel I times had to be multiplied by a factor of 1.05 to convert them to "real" time. We have not made this additional correction but instead have expressed all of our results in equivalent Channel I tape recorder time units. Table C-l shows the measured time intervals between key transmissions on Channel II and the cumulative time, measured from "hold everything" at which these transmissions occurred, corrected for the speed difference between Channel IIA and Channel IIB.


c. Repeats

Five repeats are evident from listening to the tape. All of these occurred in places where there were distinctive audible transmissions. As explained below, not all of these repeats actually increased the duration of the tape. In two cases the stylus apparently jumped backward to the previous track and started to repeat, but then jumped forward to the correct track before the recorder completed a single revolution. When this occurred, the duration of the tape would not have been lengthened relative to the duration of the original record. The strip charts of Figures C-1 and C-2 provide the detailed information from which we can determine whether the duration of each repeat was an integer multiple of the period of rotation of the record or not, and we used them to identify these two cases.

The three repeats that can be unambiguously identified by listening and by examining the strip chart pattern are:


Repeat 1 at 65 sec. 6.2 sec. added, 6.6 sec. corrected time
Repeat 3 at 122 sec. 3.8 sec. added, 4.0 sec. corrected time
Repeat 6 at 177 sec. 3.5 sec. added, 3.5 sec. corrected time


The repeats are multiples of about 3.5 seconds (corrected time), which time can be taken as the period of rotation of the recorder (the angular velocity of the recording disk on the Gray Audograph is not constant).

The strip charts also can be used to measure accurately the duration of the silences. We found one very long (7 second) silence, starting at 155 seconds that we believe was caused by a repeat during a portion of the tape in which there were no distinctive audible patterns. Therefore, we have:


Repeat 5 at 158 sec. 3.3 sec. added, 3.5 sec. corrected time


All of these repeats caused the Channel II times to be increased and the tape timings must be reduced to correct for them. This is done in Table C-1 in the columns labeled, "Cumulative Time." There are two other possible repeats, one at 96 seconds (repeat 2) and the other at 129 seconds (repeat 4). The first of these is not a repeat that caused the tape to be lengthened, since only a single word (notified) of a longer passage is repeated. The second, repeat 4, is less clear:


Repeat 4 at 129 secs 2.3 secs added, 2.4 secs corrected time


Note that this is not a multiple of 3.5 seconds. In the case of this repeat and repeat 2, the stylus apparently jumped back for a fraction of a revolution and then skipped forward to the correct track, thereby terminating the repeat. The fact that neither of these lasted a complete rotation means that there was not a spurious increase in the tape duration, and the timings should not be corrected.


d. Silences

We are told by James Bowles that the recorders had hold relays which kept them on for approximately 4 seconds after a transmission ended (the time between the end of a transmission and the recorder turnoff depends on sound intensity and is longer for very loud sounds). We do not know the threshold for this hold relay, but it is reasonable to assume that it was about 10 db below the peak signal voltage.

If a silence is less than 4 seconds, the recorder would not stop and the recorder time would correspond approximately to real time. If a silence is longer than 4 seconds, the recorder would stop and there is no simple way of determining the duration of the pause that might have occurred before it restarted.

Note that starting with "Go to the hospital" at zero seconds to silence A at 106 seconds, all silences are less than 4 seconds. The Channel II recorder must have run continuously during this interval. Starting at 106 seconds, we have a number of silences greater than 4.5 seconds during which the recorder could have paused. They are:


Silence A at 106 sec. 4.9 sec. duration, 5.2 sec. corrected time
Silence B at 132 sec. 4.5 sec. duration, 4.8 sec. corrected time
Silence C at 145 sec. 5.0 sec. duration, 5.3 sec. corrected time
Silence D at 162 sec. 5.5 sec. duration, 5.6 sec. corrected time
Silence E at 189 sec. 5.5 sec. duration, 5.5 sec. corrected time


This pattern of pauses means that, although the tape ran continuously for the first 106 seconds, during the second 100 seconds it apparently paused 5 times. During any of these pauses an indeterminate amount of time could have passed before the recorder restarted.

If during these 5 pauses the recorder had stopped for a total of 46 seconds, the "hold everything . . ." transmissions on the two channels would have coincided with time. We have no data that would allow us to determine how long the recorder actually stopped. It does not seem unreasonable that there would have been 46 seconds that Channel II was not being used during the period that the motorcade was occupied with making the trip to Parkland Hospital at high speed. In Appendix D, definite evidence is given that the Channel II recorder made at least one stop of 2.9 seconds duration between "hold everything . . ." and "You want . . . Stemmons."



From Table C-1 we see that:

1) On Channel I, "hold everything . . ." (which coincides in time with the last of the BBN "shots") occurs 171 seconds before "you want me . . . Stemmons"

2) On Channel II, "go to the hospital" occurred 189 seconds before "you want me . . . Stemmons" and 64 seconds before "hold everything . . ."

By this analysis, the last of the BRSW "shots" occurred at least 18 seconds after Chief Curry issued his instructions, "Go to the hospital," and the events identified by BRSW/WA could not have been the shots of the assassination. Except for determining the correction factor for time measurements in Channel IIA, this result does not require that the two "hold everything . . ." transmissions be identical; it requires only that the two "You want me . . . Stemmons" transmissions be the same. Note further that this result is deterministic, not based on probabilistic arguments. If one includes the known 2.9 second stop of the Channel II recorder discussed above and in Appendix D, the last of the impulses attributed to shots occurred at least 20.9 seconds after "Go to the hospital."

For the two "hold everything . . ." transmissions to coincide the recorder would have had to be inactive for 46 seconds, in which case the conjectured shots would have occurred at least 64 seconds after the chief's instructions, "Go to the hospital." There were five places where the recorder could have stopped, during which the 46 seconds of inactive time could have accumulated. For the events identified as shots by BRSW/WA to have occurred before Chief Curry's instructions, "Go to the hospital," at least 20.9 seconds would have to be deleted from Channel II, or added to Channel I. We see no evidence of anything that would allow us to shorten the Channel II times more than has already been done. Possible mechanisms that might permit us to lengthen Channel I are backward skips in the original Dictaphone recording of Channel I, or forward skips on playback. Backward skips on recording would require manual resetting of the recording stylus, an unlikely event given the automatic operation of the logging recorders, and would result in a superposition of recordings as discussed in Appendix D. Physical examination of the Dictabelt revealed no evidence of superposed recordings.


C-2. Analysis of Tapes Made Directly from Original Records

After the preceding analysis of the Bowles tape recordings had been completed, the Committee obtained access to the original Gray Audograph and Dictaphone recordings from the Department of Justice. These were transcribed onto tapes carefully so as to keep the amount of 60 Hz hum and other artifacts added to the tapes to a minimum.


Channel I Recording

The Dictabelt (Channel I) was transcribed using a Dictaphone playback unit, with its playback speed adjusted to be equal to the original recording speed. The 60 Hz hum from the original record was used to make this adjustment. The Dictabelt was in poor condition and it was difficult to measure accurately the period of the 60 Hz hum required for the speed adjustment. No skips or repeats were apparent in the process of transcription, nor are there indications of any on the resulting tapes. The time between the "You want me . . . Stemmons" transmission and the "Hold everything . . ." transmission, which coincided with the part of the tape where BRSW/WA said they found shots, was found to be 178 seconds. This compares with 171 seconds in the analysis of the Bowles tapes, in which we did not attempt to correct the times to real time.


Channel II Recording

The Gray Audograph disk (Channel II) could not be played on an original Gray playback unit without introducing skips and repeats. It was possible to play it successfully without either of these artifacts being introduced by using a phonograph turntable and phonograph arm, cartridge, and stylus. However, phonograph turntables operate at a constant rpm, whereas the Gray equipment maintains a constant linear velocity of the record relative to the stylus. Moreover, the Gray Audograph records from the inside out, whereas normal records begin at the outside. Thus, when the tapes are played back, there is a speed distortion that causes material at the beginning of the tape (the inside of the record) to be slowed down (time intervals between events are longer and the frequencies are lower than those originally recorded) and material at the end of the tape (end of the record) to be speeded up relative to true speed.

We were able to use the 60 Hz hum present on the tape to correct for this speed distortion. The hum level on the original record was fairly high and is easily discernible in the tapes during the many intervals of relative silence. By measuring the period of the hum at different points on the tape, we can determine the correction factor that must be applied to time measurements to convert them to real time.

The correction factor measurements for many points in the part of the tape of interest to us are plotted in Figure C-3. Note that in the interval between "Go to the hospital" at 22 seconds and "You want me . . . Stemmons" at 238 seconds the correction factor varies linearly with time. It has a value of about 0.95 at 130 seconds, the midpoint between these two events. We can relate the corrected (real) time, tc, to the measured time, tm, by

dtc = Kodtm + K'tmdtm

where Ko is the time correction factor at the midpoint and K' is the slope of the correction factor line from Figure C-3. If the midpoint is taken as the time origin and this equation is integrated over the interval "Go" to "You," we obtain

Tc,You - Tc,Go = Ko(Tm,You - Tm,Go) + K'(T2m,You - T2m,Go)/2

where Ko = 0.95 and K' the slope of the regression line in Figure C-3 is 0.0005. Since the T2m,Go = T2m,You, given that the time origin is midway between them, the second term on the right is zero, and

Tc,You - Tc,Go = 0.95(Tm,You - Tm,Go).

Substituting the values for Tm,You and Tm,Go, we find that

Tc,You - Tc,Go = 206 seconds.



By this analysis Chief Curry instructed the motorcade to go to the hospital at least 206 seconds before the "You want me . . . Stemmons" transmission. The events identified by BRSW/WA as shots occurred 178 seconds before the "You want me . . .Stemmons" transmission, or at least 28 seconds after Chief Curry instructed the motorcade to "Go to the hospital." This is a lower bound on the interval, because Channel II was sound-operated and halted when there were long periods of quiet. This second analysis confirms the findings from the Bowles tapes that the events identified by BRSW/WA as shots could not have been the assassination shots. If one includes the known 2.9-second stop of the Channel II recorder that is discussed in Appendix D, the impulses attributed to shots occurred at least 30.9 seconds after the instruction, "Go to the hospital."

The two sets of measurements are in reasonable agreement. The two Channel I times, 171 and 178 seconds (original record), show that the Bowles tapes played back about 4% faster than real time. If we apply this same correction factor to the Channel II time obtained from the Bowles tape, we obtain 197 seconds as the estimate of the elapsed time between "Go to the hospital" and "You want me . . . Stemmons." This compares with 206 seconds obtained from the tapes made directly from the original records. The difference, only 9 seconds, is probably due to the artifacts of the Bowles tapes: undetected skips, a sequence interpreted incorrectly as a repeat, or too low an estimate of the gap duration. We tried to be conservative in correcting for the artifacts on Channel II of the Bowles tapes and it is not surprising that the time interval between "Go" and "You" obtained from the Bowles tape is smaller than that obtained from the tape of the original recording. The tapes from the original records have fewer artifacts and a more certain history. They are believed to provide more accurate estimates of the time intervals than the Bowles tapes.


Back to the top



Table of Contents


Back to archive of acoustical evidence

Back to JFK reports and documents menu

Back to JFK menu

Dave Reitzes home page



1. Appendix to Hearings Before the Select Committee on Assassinations of the House of Representatives Ninety-Fifth Congress, Volume VIII, US Government Printing Office, Washington, DC, 1979.

2. James C. Bowles, The Kennedy Assassination Tapes: A Rebuttal to the Acoustical Evidence Theory (copyrighted and unpublished).

3. 3. Hearings Before the President's Commission on the Assassination of President Kennedy, US Government Printing Office, Washington, DC, 1964.

4. Report released December 1, 1980, by the Federal Bureau of Investigation and prepared by the FBI Technical Services Division, Washington, DC, and dated November 19, 1980.

5. Minitab Manual, by Thomas A. Ryan, Jr., Brian J. Jainer, and Barbara F. Ryan, published by Minitab Project, Statistics Department, 215 Pond Laboratory, Pennsylvania State University, University Park, PA 16802.


Back to the top



Table of Contents


Search this site
    powered by FreeFind

Back to archive of acoustical evidence

Back to JFK reports and documents menu

Back to JFK menu

Dave Reitzes home page