I. INTRODUCTION

On the day of the President Kennedy assassination, two Dallas Police Radio Channels were active and communications on these were routinely recorded. These channels were designated as Channel I and Channel II. Channel II was assigned to communications related to the presidential motorcade and was embossed on a Gray Audograph plastic disk. Channel I was to handle other routine Dallas Police radio communications and was embossed on a Dictaphone belt recorder model A2TC. Both recorders were sound-activated. At the time of the assassination, a microphone on a police motorcycle was stuck in the open position and transmitted continuously for about five minutes on Channel I, making a recording on the belt recorder.

The House Select Committee on Assassinations suspected that the motorcycle with the stuck-open mike may have been in the presidential motorcade and the officer may have incorrectly tuned his radio to Channel I. If this was indeed the case, the stuck-open mike may have picked up the gun-shot sounds, which would have been recorded. Although no shots could be heard on the recording, an acoustic analysis of the recording might reveal information as to the number of shots fired and their direction.

At the request of the House Select Committee on Assassinations, the Dictabelt recording was studied by James Barger, Scott Robinson, Edward Schmidt and Jared Wolf (BRSW) of Bolt Beranek and Newman Inc., and later by Mark Weiss and Ernest Aschkenasy (WA) of Queens College. In an initial report to the committee on Sep. 11, 1978, and in a later report in January 1979, BRSW concluded that the recording contains four sounds, which they attributed to probable gun shots and that with a probability of 50%, the third was due to a shot from the grassy knoll area of Dealey Plaza. Later WA studied the echo pattern of the "third shot" analytically and their conclusion was that with a 95% probability it represented a shot from the grassy knoll area. However their timings differed from BRSW timings by about 200 ms. BRSW subsequently reviewed WA results and agreed with their findings. This conclusion, together with the known shots from the Texas School Book depository, was the basis of the finding by the House Select Committee on Assassinations that "scientific acoustical evidence establishes a high probability that two gunmen fired at President John F. Kennedy."

On December 1, 1980, the Federal Bureau of Investigation (FBI) released a report, prepared by its Technical Services Division and dated November 19, 1980, with the findings that the above conclusion of the House Select Committee on Assassinations was not valid and that the acoustical evidence presented "did not scientifically prove that the Dictabelt recording on Channel 1 . . . contains the sounds of gunshots or any other sounds originating in Dealey Plaza. . . ."

The Committee on Ballistic Acoustics was established by the National Research Council in the fall of 1980 in response to a request from the Department of Justice for a review of the methodology employed in the evaluations of the recorded acoustic data and the conclusions about the existence of a shot from the grassy knoll. One of the authors of this report (RLG) was a member of this committee and the other two authors have collaborated with him in the investigation reported in this paper.

In the first months of its existence the Committee studied the analytical techniques used by BRSW/WA. As a result of these studies, Committee members, working from magnetic-tape copies of the embossed plastic recordings, found errors in the previous studies and faults of methodology. These faults were sufficiently serious that, by the end of the first Committee meeting, no member was convinced by previous analyses that there was a grassy knoll shot.

The Committee was greatly helped in its studies by the suggestion volunteered by Steve Barber, a private citizen of Mansfield, Ohio, that on Channel I overlying the relevant acoustic impulses there was an almost unintelligible voice communication which he thought was cross-talk from the Dallas Police Department Channel II, as heard on the tape copy of the Gray Audograph disk. The relevant phrase on Channel II was "hold everything secure until the homicide and other investigators can get there." We shall refer to this phrase as the "Hold-Everything" phrase. On Channel II this phrase occurs about a minute after the assassination. If the messages on the two recordings are the same and if they were recorded simultaneously, then the acoustic impulse pattern on Channel-I recording would be a minute too late to be shots. This is a very important piece of evidence and should be examined very carefully.

A natural explanation of this cross-talk is that the motorcycle with the stuck-open mike may at times have been close to another motorcycle which had its radio tuned to Channel II. Thus the stuck-open mike may have picked up parts of the Channel II communications and rebroadcast them on Channel I. If this was indeed the case, we might get a series of tie points between two radio recordings with identical "real" timings. This will of course not give absolute timings on the two channels but at least provide relative timings between the two.

The two recording machines were sound activated by the audio arriving at the DPD radio room over telephone lines from remotely located receivers. They would stop recording if the silence between communications lasted more than 4-6 seconds. During the critical minutes, the Channel-I recorder worked continuously because the transmitter with the stuck-open mike was transmitting continuously. Most of the time, it transmitted only motorcycle engine noise and other ambient noise picked up by the stuck-open mike. Occasionally it would transmit cross-talk from Channel II if a receiver tuned to Channel II was its vicinity. In any case, during those five minutes, because of the continuous recording, time differences on Channel I tape should equal "real" time differences (although the speed of neither recorder was closely controlled nor even precisely constant). On the other hand, Channel II recording was not continuous; on the tape, there are several silences of more than 4 seconds which might represent Channel-II recorder silent periods of any duration greater than 4 seconds. Thus the time differences between tie points on Channel-II tape may be less than (or at most equal to) time differences on corresponding tie points on Channel-I tape.

On Channel II, the "Hold-Everything" phrase is about 3.5 seconds long. During this part of the Channel-I recording, there is fairly heavy noise, presumably due to the motorcycle engine, and the Committee could not (by listening) confirm the presence of cross-talk in this region of the recording. Therefore, spectral analysis of the relevant portion of the recordings was done in order to confirm or deny the hypothesis of cross-talk. By visual inspection of the two spectra, 27 spectral features were identified and matched. Their relative timings and frequencies were compared. Timing and frequency independently indicated a speed difference of 6.7% between the recordings. After correcting for this speed difference, timing and frequencies of these features on both the recordings matched with good accuracy. Since feature matching is a subjective process, a more objective experiment of cross-correlating the two spectra was carried out at IBM. This experiment is described in more detail in the next section.

Another instance of cross-talk is present in the phrase "You want me to still hold this traffic on Stemmons until we find out something or let it go?" ("Stemmons" phrase). For this phrase, just by listening alone, the cross-talk is clear and the two spectra match very well (after correction for the difference in recording speeds on the two channels). This phrase is about 4.5 seconds long and occurs 170 seconds after the "Hold-Everything" phrase on Channel I. On Channel II, the time difference is only 125 seconds. This difference of 45 seconds is, as explained above, due to the Channel-II recorder stopping during several silences.