-6- 
launcher. For each probe two measurements were carried out each in the range between 
1 °C and 5 °C, 10 °C and 11 °C, and 19 °C and 20 °C, and were compared against 
the PtlOO readings recorded. 36 Sippican Deep-Blue probes were calibrated according 
to this procedure, one box in 1988 and two boxes in 1989. All probes were purchased 
about 6 months prior to the test. In addition, 6 probes XBT-7 manufactured by Sparton 
of Canada Ltd. were calibrated in 1989. 
Except for three probes, all error functions /\T calculated are significant within a 95 % 
confidence interval due to a linear fit and show a positive derivation (Fig. 3a-d). The 
considerable deviation of regression lines within each batch indicates a better thermistor 
accuracy than that given by Sippican as ± 0.2 K. The positive derivation, however, is 
assumed to be caused by the Bathy System controller. This assumption is supported by 
a study published by Roemmich and Cornuelle (1987) who calibrated 24 "Deep Blue" 
probes without finding a systematic derivation. For their study they used a self 
constructed XBT digitizer built in a PC-type microcomputer. According to the results 
presented a systematic error in the order of + 0.2 K can be expected for measurements 
in the tropical mixed layer. Fortunately, this bias is not relevant in the North Atlantic 
(T < 20 °C) and hence is no serious problem for the quality of our SOOP data. 
A week after the laboratory tests in 1989, 12 calibrated probes were used for an in-situ 
comparison with two CTDs in the Norwegian Trench (water depth 700 m). An 
autonomous SIS CTD, with a self-adjusting high resolution pressure sensor, was mounted 
at a NB Mklll CTD. The sensor systems of both instruments were horizontally 30 cm 
apart. No significant deviation between the profiles of the two CTDs could be observed. 
First results of this comparison are shown in Fig. 4a-c. The XBT traces are corrected 
according to the calibration results. The depth difference J\z = z(XBT) - z(CTD) of 
8 XBT traces are negative, whereas that of 3 traces match the CTD trace (solid line). 
1 probe seems to have failed (# 9). The clear positive temperature increase in the upper 
part of trace # 9 may be indicative of the so-called "bowing" (IOC, 1989). A gradual 
increase, or "bowing", in temperature was observed in the mixed layer only and first 
reported by Bailey et al. (1989). In the segment of the strongest gradient, which is 
located between 290 m and 440 m, the maximum negative depth difference to be found 
is about 14 m (# 10, # 12) at the lower end of this segment. The depth difference of