44 THE WATER 



Negretti and Zambra, of London, and a modified 

 instrument, made by Richter, of Berlin. 



Fig. 8 shows the Negretti and Zambra instrument 

 as it is sent down, and Fig. 9 the thermometer by 

 itself reversed. The thermometer is fixed in a frame 

 which can be turned over at will. On reversal either 

 by a messenger (as in Fig. 8), or by a propeller, the 

 mercury thread breaks at the constriction, and fills 

 the small bulb at the end of the capillary and also part 

 of the capillary itself. It is read in the reversed 

 position, and, apart from certain corrections, to be 

 mentioned afterwards, the graduation on the stem 

 gives the desired temperature. The thermometer 

 should remain at the required depth for about five 

 minutes to take the temperature of the water. 



There are three sources of error, which all arise from 

 faults of construction. In the first place the whole 

 of the mercury in the main bulb may pass the con- 

 striction on reversal. This is not a common fault, 

 and thermometers subject to it should be discarded, 

 though they may at times work well. Thermometers 

 which show no signs of this defect on reversing in the 

 hand, may fail when subjected to the vibration caused 

 by hauling in from a great depth, especially in a rough 

 sea. They should therefore be tested by reversing 

 and tapping on a table. A second cause of inaccuracy 

 is due to the mercury which remains in the bulb 

 expanding past the constriction, when the ther- 

 mometer is hauled up through warmer layers, or 

 brought into a -warmer atmosphere. The S -shaped 

 dilatation, which differs slightly in the two models, is 

 intended to catch this mercury, but it fails to do so at 

 times. 



The third and most serious defect is that the mercury 

 does not always part exactly at the constriction, and, 



