51 



construction recommended in United States Bureau of Standards 

 Kesearch Paper 223. All switch contacts have been improved, the 

 size of the electrolytic cells has been reduced so that the resulting 

 volume is about 35 ml per cell, five cells for samples have been pro- 

 vided, the telephone receivers are adjustable, and all sides, top, and 

 bottom are removable panels, the instrument being assembled on an 

 angle-brass frame to wliich the removable panels are attached. One 

 important respect in which the new model differs from the earlier 

 instruments is that the slide wire of the old models is replaced by a 

 series of dial switches. The circulating medium in the constant 

 temperature bath is oil and advantage has been taken of this feature 

 to improve cell connections which are now made under oil, the cell 

 selector-switch also being under oil. All wiring is in the oil bath, 

 in conduits, or back of bakelite panels, thus being much better pro- 

 tected against the accidental spillage of sea water. The design of the 

 oil circulating system has been improved both in quantity and 

 direction. 



After the salinity bridge was assembled and installed the thermo- 

 stat was adjusted to regulate the temperature of the oil bath at 

 22.5° C, a value chosen as being just high enough to include the maxi- 

 mum probable room temperature of the laboratory. A test run was 

 then made to determine the thermal characteristics of the instrument. 

 The sample-bottles, after being brought to the laboratory, are usually 

 stored on the deck of the laboratory until they are measured. In 

 this position they most frequently have a temperature of about 14° C. 

 when it is time for them to be measured. Therefore it was considered 

 that a fair test would consist of introducing into the bridge a sample 

 at a temperature of 12° C. and then balancing the bridge at successive 

 1-minute intervals until constant conditions were attained. This 

 was done, the test being continued for 34 minutes although no change 

 could be detected after 28 minutes. The measurements fell on a 

 remarkably smooth curve from which the following results were 

 obtained: 



From the above it will be seen that the bridge passed this test 

 satisfactorily and it was concluded that 17^ minutes would be suflB- 

 cient length of time to allow between the introduction of a sample and 

 its measurement. As there were five cells for samples this meant an 

 average speed of about 4 minutes per measurement. 



