National Academy of Sciences (NAS)

Report of the Committee on
Ballistic Acoustics






Washington, DC 1982




The Committee considered seriously the possibility that the impulses analyzed by BRSW/WA might have been overlaid at a later time by the "hold everything . . ." message; if that had occurred the presence of the message would not automatically imply that the impulses occurred well after the assassination. The Committee examined four kinds of evidence on the possibility: (1) physical evidence concerning overrecording or Dictabelt substitution, (2) likelihood of possible scenarios for overrecording, (3) compatibility of "hold everything . . ." timing with other firmly established evidence, and (4) recorded acoustic evidence on the recordings such as the characteristics which show that the recorded cross talk passed through a radio receiver with automatic gain control (AGC). These four kinds of evidence are discussed below.

(1) Physical evidence concerning overrecording or Dictabelt substitution

The original DPD Channel I Dictabelt was examined by some Committee members and by William Sturtevant of the Dictaphone Corporation for possible physical evidence of a double recording. No indications of superposed recordings were found. The track in the region of the "hold everything " expression was examined with particular care. The tracks were remarkably clear and parallel and showed no indications of superposed recordings.

There was considerable writing with a china marking pencil on the surface of the Dictabelt. The markings give in one handwriting "11-22-63, PL2" and an encircled "10." The times 12 5 and 12 40 in a different handwriting also appear as do the letters J and H. These markings were similar to those on the other Dictabelts made that day. A photograph of the Dictabelt has been submitted to Ms. Doris Schwartz, who serviced the recorder during the period in question and who now lives in Duncanville, Texas. Although Ms. Schwartz does not recognize the other handwriting, she does identify the "11-22-63, PL2, 10" as her own handwriting. She uses an unmistakable 2 and feels that the specimen is the original belt. With this identification, any substitution of a different Dictabelt for the original could not have been accidental; it would have had to be deliberate, and it would have involved counterfeiting the writing for the date and belt number.

(2) The likelihood of scenarios for over recording

Although there is good reason to believe that the belt examined was the original one, the Committee also considered the possibilities that the belt examined might itself have been a later copy of an earlier Dictabelt with a double recording, in which case the physical examination would not show evidence of superposed recordings, or might have been a copy substituted for the original with the "hold everything . . ." superposed while copying. To explain in one of these ways the presence of the "hold everything " phrase except by deliberate intent would require an incredible array of accidents, each of which is highly unlikely. For such accidents to have occurred immediately following the assassination, the Dictabelt would have had to be knocked backward by just one minute of recording time, this unprecedented accident would have had to occur within the first minute following the assassination, it would have had to occur in a manner to leave no acoustic evidence, and in addition someone irresponsibly would have had to copy the Dictabelt, counterfeit the identification writing, substitute the new Dictabelt for the old one, and throw away the original, despite both the importance of the case and the well known police requirement not to destroy evidence.

Comparable difficulties confront the hypothesis that there was an accidental superposed recording at a later time. In that case, the Dictaphone playing potential evidence of a presidential assassination would have had to be carelessly set to record instead of listen, and a recording of Channel II at the same time would have had to be playing accidentally in the background. In addition the Channel II playing would have had to be set accidentally to the very narrow range that could indicate incompatibility; if the setting had occurred one minute earlier it would have produced no problem and one minute later it would have been detectable as an impossible timing. The double recording would have had to continue at least two minutes to account for the cross talk discussed in Appendix C. Further, there would have had to be an irresponsible substitution of a copied Dictabelt, with counterfeited written identification and a subsequent loss of the original, to obscure physical evidence of a double recording. For all this to occur by accident again seems highly unlikely. Similar difficulties face the hypothesis that the original Dictabelt was copied to another Dictabelt, that the cross talk was picked up during the copying, that the second Dictabelt was accidentally substituted for the original, and the original was destroyed.

There are comparable difficulties in attributing the presence of the "Hold everything . . ." message to a deliberate superposed recording and a deliberate effort to conceal this action. It is particularly difficult to believe that within the first minute after the assassination someone deliberately knocked backward the Dictabelt to confuse the interpretations of the transmission from the open microphone. No one knew in advance, or even at the time, that there was to be an accidentally open microphone or where it might be located. No one could tell in advance how much to displace the Dictabelt to assure that the next transmission of cross talk between Channels I and II would be exactly superposed on the assassination. No one knew in advance there were impulses that would years later be interpreted as assassination shots or that the cross talk between the two Channels would eventually be interpreted at an even later time. Anyone who substituted a copy of the tape to hide the physical evidence of a superposed recording, and who counterfeited the written identification exposed himself to a serious risk of being caught in exchange for a potential benefit that was remote and unlikely to materialize.

Similar difficulties are presented by the hypothesis that the "hold everything . . ." message was deliberately added at a later time. Since the message appears on the original Dictabelt and on a copy made by James Bowles in March 1964, the deliberate superposed recording would have had to be made before this date, long before the BRSW/WA analysis indicated there might be information worth confusing. It could not be apparent in advance that the "hold everything . . ." expression would ever be correctly interpreted. The person deliberately faking this evidence and then substituting a new recording with a counterfeit identification for the original one faced a large chance of being caught in exchange for only a very small chance that the deception would be either necessary or helpful. It seems equally unlikely that such an elaborate and risky substitution of forged evidence would have been made for any other reason, such as to cover up a broadcast of an embarrassing remark.

In summary, all the possible scenarios by which a copied Dictabelt with a forged identification could have been substituted for the original seem highly unlikely.

(3) Compatibility of "hold everything . . ." timing with other firmly established evidence

Since the "hold everything . . ." expression on Channel II occurs approximately one minute after the expression "Go to the hospital . . . ," hence, more than one minute after the assassination, and since the Channel I recorder ran continuously during much of the subsequent time, the cross correlation between the two channels has implications for the timing of some other events following the assassination. Some of this other evidence is unreliable, but some is firmly established. The Committee examined such evidence to be sure there was no incompatibility between the timing sequence implied by the "hold everything . . ." identification and the firm limits established by other evidence. Some of these limits are discussed in Appendix C where it is noted that later cross talk between the two channels is compatible with this identification.

A quite different test is implied by the sounds of sirens, as discussed in Appendix E, where it is shown that the time interval between the assassination and the appearance of siren sounds is compatible with a reasonable speed of the motorcade to the hospital, provided that the microphone picking up the siren sounds is at the Trade Mart or some other point closer to the hospital.

Finally, the time announcements on Channel I provide further support to the assumption that there was no double recording. The last time announcement on Channel I before the conjectured shots and the "hold everything . . ." statement was "12:28" and the first time announcement thereafter was "12:34," a nominal difference of six minutes. When the Channel I recording is run at a speed that reproduces the recorded hum at 60 Hz, the elapsed time between these two announcements is 389 seconds or 6 minutes 29 seconds in full agreement with the six minute nominal difference, since the time announcements were at irregular intervals and never included seconds. If the recording head had been moved back by 64 seconds of recording time to superpose a later "hold everything . . ." on top of earlier recorded acoustic impulses, the elapsed time between the 12:28 and 12:34 time announcements would have been 453 seconds or 7 minutes 33 seconds in disagreement with the time announcements. It should be noted, however, that the time announcements were not highly reliable since they were obtained from a variety of different time indicators, so this evidence alone while being supportive of the Committee's position, should not be regarded as conclusive. It should also be noted that by timing measurements over a wider range, the BRSW report, page 33, concludes that the Channel I recorder "operated continuously"; a 64-second recording overlap would be a marked discontinuity. If it were argued that a superposed recording occurred by a Dictabelt displacement which corrected itself before the 12:34 time announcement, a groove cutting at a different angle and spacing should have been observable, and no such groove was seen during the physical examination of the Dictabelt. Furthermore, such a scenario would leave intact the evidence from Appendix C that the relevant acoustic impulses occurred well after the assassination.

(4) Recorded acoustic evidence on the Dictabelt, such as characteristics which show that the recorded cross talk passed through a radio receiver with automatic gain control (AGC)

From Figures B-4, B-5 and B-6, it is apparent that there are heterodyne tones present on Channel I which do not exist on Channel II. In fact, these tones are the key to the demonstration that the underlying Channel II sounds present in Channel I were recorded via radio and not at some later time.

Close attention to these Channel I heterodyne tones which are not present in Channel II at the same time, shows that this tone is followed immediately (over a period of about 30-50 milliseconds) by a rapid suppression of the Channel II signal, which then assumes a constant value for the duration of the tone. At the end of the Channel I tone, the Channel II transmission rises over a period of 100-200 milliseconds to its original value. This is to be understood if the Channel I tones are heterodynes caused by additional transmitters appearing on the Channel I band, and dominating the open-mike transmission. Thus, the radio receiver will readjust its automatic gain control (AGC) to maintain a steady (intermediate-frequency) signal level in its output. (See the heterodyne beginning at "A" in Channel I on Figure B-4.) This is natural for such radio receivers. At the end of the presence of the strong carrier signal, the AGC readjusts to provide a steady output level for the open-mike transmission. Hence, any probing tone on the microphone will suffer this same variation due to the AGC action in the presence of a sudden interfering carrier on Channel I.

In many cases, however, there is present in the Channel I output a brief tone, of strength similar to those noted in the previous paragraph, which is present also in Channel II. These brief tones in no case show a change in level at the beginning and end. (See the Channel II heterodyne marked "C" on Fig. B-4.) This is to be understood because they enter the open microphone and are broadcast at a steady carrier level on Channel I. Whatever behavior there was of AGC on Channel II is reflected in the level of the heterodyne on Channel II, and there should be no further distortion of this level by action of AGC on Channel I.

This analysis demonstrates that cross talk from Channel II was directly recorded onto a Channel I recorder through a radio receiver and not at some later time coupled either by electrical or acoustical coupling into a recording of Channel I. Had this latter taken place, there would have been no difference in AGC action on the heterodynes recorded only on Channel I from that of the brief tones present in Channel II as well. This is true, independent of any assumption that might be made of automatic level control on subsequent recordings.

The crucial evidence within the "hold everything . . ." interval itself occurs in a 0.5 seconds region of the Channel I recording shown in Figure B-6. This corresponds to Channel II times of 32.00-32.50 seconds, aligned with the corresponding signals on Channel I. Note that the Channel II recording is the recent one made by this Committee from the gray Audograph disk, without "repeats." The suppression of a Channel II heterodyne is seen in Figure B-6 and has been demonstrated quantitatively by the digital data plotted in Figures B-1 and B-8.

Although the above evidence on the recordings shows conclusively that the "hold everything . . ." and other cross talk phrases were recorded through radio communication and not by later copying, there is other acoustic evidence on the recordings as well that supports the conclusion that the Channel II cross talk was recorded on Channel I at the actual time and not in later copying. The phrases "You want me . . ." and "I'll check all these motor cycle radios" present on Channel I are derived from Channel II. However, on Channel I, the time difference between the beginning of these two phrases is 15.10 seconds, while on Channel II it is 12.80 seconds -- a difference of 2.9 seconds. This time difference can not be due to a forward skip of Channel II on playback or a backward repeat on record since 2.9 seconds is not an integral multiple of the 3.58 seconds Audograph turntable period at this time, as measured by observable print through on the recording. If the Channel I recording is made at the actual time, this difference is easily attributed to the sound actuated Channel II recorder having stopped for 2.9 seconds between these two phrases, while the Channel I open microphone recording continued on without interruption. This accounts naturally for the 2.9-second difference; but if the Channel II recording had been later overlaid onto Channel I the timing of the two would normally have been identical, not differing by 2.9 seconds. The timing analysis was made between tapes for which the best warp between "You want me . . ." on the two tapes was 0.988 while that for "I'll check all these . . ." was 0.991.

Finally, if one wishes to criticize the "hold everything . . ." evidence on the grounds that the Dictabelt is a copy and not the authentic original Dictabelt, then he must also recognize that all the acoustic evidence interpreted as favoring gunshots also comes from the same unauthentic source.

As a result of all these considerations, including especially the ones showing that the cross talk passed through a radio receiver with AGC, the Committee determined that there was conclusive evidence that the "hold everything . . ." expression was recorded on Channel I at the same time as on Channel II and that the acoustic impulses attributed to gunshots were recorded well after the President was shot and the motorcade had been instructed to go to the hospital.


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