and differences were noted. Because normal system depth errors of ±2% or ±15 feet can 

 produce resulting apparent temperature errors at depths located within rapid temperature 

 changes, care must be used to select temperatures within a mixed layer. * Typically, 

 Sippican would expect up to .4°C worst case temperature disagreement between simul- 

 taneous probe drops in the same constant temperature body of water, exclusive of depth- 

 induced temperature errors. 



Of the trace results in relatively constant temperature water, 76 pairs showed 

 essential agreement to within .4°C. Trace 47 (Figure 43 A) is an example of a marginal 

 failure to this criteria, where one appears to measure about .2°C low and the other about 

 .2°C high. The total acceptable probe pair drops expressed as a percent is 76/96 X 100 

 = 79.2%, statistically close to the expected value for a 90% reliable probe of (.9 X .9) 81%. 



It should be expected that the great majority of failures will occur in an upscale 

 direction (as was noted by the author) if the failure is caused by poor insulation quality 

 and will become more noticeable with depth as more insulation is exposed. 



(9) Out of 15 probe drops made in water of known temperature profile, only 13 

 provided visually acceptable data to 400 m and of the 13, only 3 provided temperature 

 data to the "accuracy criteria" at each of three monitored depths at 200, 300, and 

 400 m.** 



Comment: 



Assuming Drop 1 28D (Figure 48) is a typical example of profiles failing at all 

 three depths, the following conclusions may be drawn. 



Based on knowledge of expected depth-temperature results from previous drops, 

 this profile exhibits a consistent upscale reading with visible "rounding" in the surface 

 mixed layer, indicating possible excess leakage in either wire, probe, or launcher. Since 

 some earlier and later profiles did not exhibit his anomaly, it is concluded that the error 

 can be probably attributed to poor wire or thermistor insulation. Because the failure is 

 not catastrophic but rather marginal, insulation degradation with age is suspect. 



The following discussion relates to "Recommendations" suggested by the author. 



• Make routine surface measurements as a check on XBT system. 



A preferred method would seem to be more extensive use of A2A test canister 

 and A4 test box to detect system maintenance and calibration problems. 



• Log I/D of probe carton. 



Excellent recommendation. Questions relating to "bad" 460 m probes (and up 

 to 10% can fall in this category) versus "out-of-warranty" probes could be immediately 

 settled if this were adopted. 



• Contact Sippican when system malfunctions for analysis and repair. 



Excellent suggestion. Questions relating to launcher leakage and recorder 

 performance can be settled much more expeditiously. 



*See author's note C. 

 ''*See author's note A concerning accuracy criteria. 



139 



