but in a few instances much shorter-period waves were propagated over a few thousand
kilometers at these same speeds. The maximum speed of travel, 335 m/sec, was
roughly equal to the speed of sound at ground level.
Travel speeds for direct waves on standard equipment showed somewhat greater

variability than did the speeds for Ivy.
7.2.4 Azimuth Errors.

For distances less than 12,000 km from the explosionsite,

the maximum observed azimuth error was 11.5 degrees, and the average error was 3.2

degrees. At longer distances much~-larger errors were reported. No consistent pattern
of azimuth errors was observed that could be related to the direction the acoustic wave
travels from the source.
Azimuth errors observed for Castle were consistent with those observed on previous
tests. Errors in the azimuths computed for the dispersive train were roughly the same
as the errors for later portions of the wavetrain.
7.2.8 Yield. Attempts have been made to relate various characteristics of acoustic
signals atgreat distances to the total cnergy released by the nuclear explosion. Critical
dependence of sigual amplitude on the variable temperature and wind structure in the
upper atmosphere, coupled with difficuities in the accurate measurement of amplitude
led to a search for more-reliable indicators of yield. A possible connection between
signal frequency and yield involving a cube-law relationship based upon general scaling
considerationa was postulated. This cube-law relationship between the duration of the
first negative pulse and yield was verified for acoustic records at ranges of 7 to 600
mies from explosions at the Nevada TestSite.
A critical examination of a great many acoustic recordings at distances greater than
1,000 km from explosions in the yield range of from 1 to 500 kt led to the use of the visually observed signal periods in the vicinity of maximum amplitude for standard recordings as the best indicator of yield. For each shot, periods from selected stations were
averaged anc the averages were plotted. Similar periods were seiected from standard
recordings of the direct wave from the megaton shots of Ivy and Castle. A best powerlaw curve was computed by the mcthod of least squares for data up to and including yields
of 500 kt. This curve indicated the yield to be equal to a constant multiplied by the period
raised to roughly the third power.
Data for yields above about 100 kt fell along a curve of different slope from that for
fower yields. The best curve in this region Indicated that for megaton shots the yield
would be equal to a constant multiplied by the period ‘at maximum amplitude, for standard equipment) raised to roughly the fourtn power.
The method of measuring the period was somewhat subjective and the relationship
between yield and period very inaccurate. In addition, the method requires measurements at a number of stations for each shot in order to achieve even the semiquantitative
results noted here.
Very-large errors are inherent in this method of determining yield from acoustic
measurements. For yields up to about 100 kt, three standard errors of estimate cover
yields as smal! as a fifth and as large as five times the correct value. Errors at yields
above roughly 100 kt seem slightly smaller, although a correction for the small sample
has been applied. Three standard errors cover yields as small as a third and as large
as three times the correct value at these higher yields.
Studies of the accuracy of yield determinations from the VLF recordings were being
tuade, with effort centered on measurement of amplitude for these recordings.
Many other general indicators of yield were apparent: the existence of a dispersive
97

Select target paragraph3