Late in the cruise, between Station 105 and the last one at 

 Station 163, a calibrated shunt was introduced into the STD conductivity 

 cell. This brought most of the low salinities on scale. These stations 

 were then done in two lowerings which were labeled S and D (shallow 

 and deep) during data reduction. 



The STD has a slow-acting analog conversion circuit to con- 

 vert electrical conductivity to salinity. Although the manufacturer 

 quotes a response time of 0.3 seconds for this circuit, it was measured 

 experimentally as 2.0 seconds. This slow response caused troubles, 

 as will be shown below. The marginal ice zone is full of sharp tempera- 

 ture gradients, some as great as 7°C per meter. The winch lowering 

 speed was basically 0.55-0.65 m/sec. Up to Station 40 the lowering 

 was stopped at fixed depth intervals, but thereafter the STD was lowered 

 and raised continuously at the above speed. 



The salinity error due to lag in the compensating thermometer 

 may be roughly estimated in the following way. When the instrument 

 passes through a temperature gradient, the error in temperature will not 

 be greater than that resulting from a temperature ramp which continues 

 for several time constants of the thermometer. The first-order thermo- 

 meter response equation is easily solved for this case and the error is 

 found equal to the product of the time constant and the time rate of change 

 of temperature. Suppose one examines the extremes, assumes 3°/m as 

 a more common temperature gradient than 7°/m and assumes a winch 

 speed of 0.6 m/sec. Then the time rate of change is 1.8°/sec and the 

 temperature lag is 3.6°. This corresponds to 3.6 o/oo salinity error. 

 Actually, such sharp gradients do not continue for several time constants 

 and the error computed above is likely to be too large by a factor of 1.5 

 to 2.0. Therefore, one expects spurious salinity spikes of a magnitude 

 approximating one to two parts per thousand in the sharper thermoclines. 

 Elsewhere the errors are much more moderate. • . 



The error resulting from a temperature ramp decays to zero when 

 the ramp ends. Therefore, an STD passing through a step-like tempera- 

 ture gradient creates an anomalous spike in the salinity, pointing in the 

 same direction as the temperature change. If the temperature profile con- 

 tains a one-sided oscillation, the salinity anomaly will be S-shaped. 



Although these salinity transients made a messy record, they 

 eventually led to little difficulty; we merely ignored them in tracing the 

 records and little serious error seems to have resulted. 



Scale changing on the STD proved to be a chore because the 

 gradients were so large that one could go through two scale changes in 

 a few seconds of lowering time. There was no low- sensitivity scale 



