provide replacement units providing the data and circumstances indicate that defective 

 probes less than two years old were responsible. 



(3) Only 37.8% of a total of 518, 460 m system drops satisfied "accuracy criteria" 

 (±.24°C)* at all three depths of 200 m, 300 m and 400 m and 19.9% failed accuracy criteria 

 at all three depths. 



Comment: 



Accuracy of the system exists in two frames of reference depth and temperature. 

 Sippican states temperature accuracy at ±.2°C (±.1°C probe and ±.1°C recorder) and depth 

 accuracy at ±15 feet or ±2%, whichever is larger. 



Any apparent temperature error at specific depth is the result of both real 

 temperature error and depth induced temperature error, if a gradient exists around that 

 depth. For this reason, Sippican would establish system accuracy by measuring temperature 

 and depth at the start and/or finish of a known zero-gradient temperature/depth feature, 

 rather than an indicated temperature at an apparent depth, to eliminate the added apparent 

 temperature error resulting actually from depth error.** 



In addition, as mentioned above, probes that are out of two-year warranty may 

 exhibit greater error than the assigned ±.1°C over the 460 m depth range because of gradual 

 wire and thermistor insulation degradation with time. 



(4) XBT temperatures and temperature-depth gradients are higher than those 

 measured by other systems.*** 



Comment: 



In general, probes older than two years may have thermistor and wire leakage 

 levels that influence temperature readings. Insufficient wire and thermistor insulation 

 resistance will produce, in a majority of cases, higher temperature readings. Furthermore, 

 as more wire becomes wetted, e.g., at greater depths, its insulation leakage becomes greater 

 and exhibits greater effects on the readings, usually in an up-scale direction. Hence, actual 

 temperature error would be expected to increase in an up-scale direction with depth for 

 out of warranty XBT probes. 



Unless the gradient being measured was positive, e.g., increasing temperature 

 with increasing depth, the effect of wire leakage would seem to be to reduce the actual 

 gradient rather than increase it. 



(5) Of 559 profiles made that reached at least 200 m when neither of the two 

 recorder systems that malfunctioned were contributing errors, 9.7% (54) exceeded average 

 hydrocast and STD/SV temperatures by ^.5°C at one or more of the sampled depths of 

 200 m, 300 m, or 400 m. 



Comment: 



A number of these units might be expected to fail because of normal production 

 tolerance on wire and thermistor insulation resistance. However, a quantity approaching 



*See author's note A. 



**See author's noteC. 



***See author's note D. 



137 



