We used an expendable bathythermograph (XBT) system manufactured 

 by General Motors. The XBT's were modified to reduce the fall rate 

 from the normal fall rate of 20 ft/sec to about 7 ft/sec. Special am.plifier 

 circuits were added to the standard XBT deck electronics so we could 

 obtain selectively expanded traces of the vertical temperature profile. We 

 used the XBT's in three ways: (1) as modified free-fall units; (2) with the 

 probe unit lowered at a nearly constant rate by hydrographic winch; and (3) 

 with the probe repeatedly lowered and raised slowly by winch through a given 

 depth interval with the electronics at high amplification (0. 023''C/in width of 

 chart paper). 



A section of a typical profile taken by method (2) is shown in Fig, 3a. 

 In the process of reproduction the trace has undergone some distortion, 

 nevertheless the stair-step nature of the profile is evident. A section of 

 a profile taken by method (3) is also shown in Fig, 3b, in this case covering 

 that portion of the smaller scale profile shaded in the figure. The profiles 

 obtained are much like stair steps in both traces. 



Four aspects of the traces in Figures 3a and 3b are especially noteworthy. 

 First, the temperature within each layer is extraordinarily homogeneous. For 

 example, the layer 6. 7 m thick, centered at 310 meters, shows less than 

 0.001°C variation. Second, the interface thickness is less than 20 cm based 

 on the measured response time of the recorder and the winch speed. The lip 

 at the end of each interface is recorder overshoot. Third, the magnitude of 

 the temperature increment is remarkably constant over the profile section 

 shown here with the most common teraperature increment averaging about 

 0.027°C. However, there also appear to be some temperature increments 

 of one-half the common average which we will refer to as half steps. One 

 such increment is encircled in Figure 3. Fourth, below 350 meters where 

 the temperature gradient is much snaaller, the thickness of the layers is much 

 larger. However, the thicker layers appear to be separated by several 

 very small transition steps. 



Our measurement plan on T-3 was largely devoted to time series 

 profiling in the upper portion of the Atlantic water. Figure 4 shows one series 

 of 22 XBT casts taken at 15 minute intervals using method (1). Also shown 

 is the first of another series of XBT's begun about 18 hours after the comple- 

 tion of the first series. Since depth was not measured, the traces have been 

 aligned along the specific discontinuity at approximately 250 meters. 



Half steps about which the traces are aligned persisted unchanged in 

 size for the duration of both series, about 24 hours. From 220 to 300 

 meters the steps are remarkably stable; the number of the steps remained 

 constant and the dimensions of the steps remained nearly constant. We 



461 



