THURAS: DENSITY MEASUREMENTS 607 



through a properly placed external circuit, an electromagnetic 

 attraction is exerted in a vertical direction on the bulb. The 

 utmost precautions had to be taken to maintain a constant 

 accurately known temperature, to avoid any jarring of the 

 apparatus, and to avoid errors due to eddy currents in the 

 liquid. 



In all of these modifications of the total immersion method of 

 measuring the density of liquids, the outstanding difficulty is the 

 production of small and accurately known variations in the 

 bouyancy of the sinker. Some easily variable, easily measur- 

 able force that will produce vertical displacements of the bobbin 

 is wanted. 



Such a force can most conveniently and easily be obtained 

 indirectly by varying the temperature of the liquid. This method 

 is employed in the apparatus that is described below. Numerous 

 analyses having showb that the composition of sea-water salts is 

 very nearly the same in all parts of the ocean and at all depths, 

 the concentration alone changing, the temperature coefficient of 

 expansion of all samples will be the same and the constant of the 

 apparatus will need be determined but once for all. 



The maximum variation in the density of the sea water in the 

 open ocean is less than one per cent, so that the density of the 

 most saline water can be made equal to the density of the least 

 saline water by raising its temperature 20°C. Also, the coeffi- 

 cient of expansion of sea water is so low that a sufficient degree of 

 accuracy in the density can be obtained without an excessively 

 close reading of the temperatures. This method also elimi- 

 nates measurements other than that of temperature, which is 

 always needed in accurate density determinations. 



The apparatus (fig. 1) consists essentially of a test tube con- 

 taining the float, or bobbin, and the liquid to be measured, a 

 stirred variable temperature bath, and an accurately calibrated 

 thermometer. 



The bobbins are made of Jena glass No. 16 m about 5 cm. long 

 and 12 mm. in diameter. They were carefully annealed about 

 eight months ago and have probably settled down to a definite 

 volume. The glass test tube to contain the liquid to be meas- 



