1906-7.] On the Application of a Differential Densimeter. 213 
length of fine-bore thermometer tubing N. Finally, the open ends of the 
tubes enter the cisterns G, H, which are placed upon the supports P, R. 
These supports were made from two stout glass jars 8 cms. high ; their 
bases and mouths were ground plane and parallel, and their heights 
adjusted until they were nearly equal. The mouths of the jars were closed 
by glass plates cemented on with marine glue. The use of blocks of wood 
or metal would be liable to introduce errors, since the former might become 
more or less distorted, and the latter corroded, by the unavoidable contact 
with sea- water; glass is free from these objections, The supports P, R 
occupy marked positions upon a slab Q, of plate-glass, cemented to the 
upper surface of a well-formed brick, which may be levelled by means of 
the set screws shewn below. 
The two cisterns were cut from Erlenmeyer flasks of Jena glass, having 
a capacity of 200 c.c. These were chosen on account of the ease with 
which two flasks having the same diameter, to within one or two tenths 
of a millimetre, may be obtained. 
The whole apparatus was secured to a stout and well-seasoned board 
of Kauri pine set up rigidly. A window, W, 10 cms. high, was cut in the 
board, and this had its centre approximately 110 cms. above the glass 
plate Q. Immediately behind and just below the window, a wooden 
bracket covered with plate-glass was fastened ; this was used for support- 
ing a travelling microscope M, to which a vertical movement could be 
imparted by a micrometer screw X, capable of indicating hOl mm. The 
carriage of the microscope rested upon, and was cemented to, a glass plate 
as recommended by Quincke. 
In order to secure uniformity of temperature for corresponding points, 
the tubes throughout their entire lengths were brought close together. For 
some experiments they were wrapped in cotton-wool, whilst for others 
they were left uncovered ; but the best plan was found to be that in which 
the tubes are jacketed with water, the jacket itself being surrounded with 
a wooden cover (fig. 2). 
The error due to any changes in the temperature of the air enclosed 
in the upper part of the apparatus is eliminated by the introduction of 
the taps t , p, which are simultaneously closed immediately before observa- 
tions of the heights of the water columns are commenced. 
The accuracy of the method is largely increased by the use of columns 
of ample length, the heights of which are determined with the micrometer. 
In measuring the lengths of the two columns of water, two only, instead 
of four readings, are required. This desirable simplification was secured by 
the use of the conical cisterns above mentioned. The diameter of these at 
