point (7, , 7.) where Z, = 4, can be computed by substituting 
T'g and 2", into the expression 
bz tT 1 
3 + 
Pop tt bye ptt 
(4 
i) 
| 
Finally, repeat until Z is close enough to the limit. 
An example of an attempt to fit the temperature data along 
the strut is given in figure 12. Interpolations of the temperature 
at half-degree intervals are given by the harmonic method and by 
the polynomial method in a rotated coordinate system. The har- 
monic series obviously gives a very bad fit; a polynomial (not 
shown) gives an even worse fit. The polynomial in the rotated 
coordinate system gives a fairly good fit. It would have been more 
instructive, though, to have calculated the depth at the temperature 
data points. 
FINAL FORM OF THE REDUCED DATA 
For figure 12, the interpolation scheme described above was 
used to determine the position of a number of isotherms relative 
to the strut. It is shown earlier how this result can be used 
to give the depth in the sea for the various isotherms. These pro- 
cesses are repeated for each sequence of temperature and pres- 
sure readings. The result may be plotted as a function of time or 
position. In figure 13, the depth of a number of isotherms is given 
as a function of time for the case of an array being carried at a 
uniform horizontal speed and at such a depth that the thermocline 
is straddled. 
Because the strut does not straddle the thermocline uniform- 
ly, the procedure in the section on interpolation does not give 
depths for all isotherms. A technique, not to be discussed here, 
has been devised to fill in isotherm gaps; these are indicated as 
dotted curves. Further analysis (i.e., auto-correlations, power 
spectra, filtering, etc.) of the reduced data is likewise not dis- 
cussed here. It should be noted, however, that the frequencies of 
29 
