THERMOMETERS 35 



investigations are being made at various depths, it is 

 better to use the deep-sea water-bottle for the surface 

 also, if the weather permits of this being done. 



Below the surface, water samples must be collected 

 by means of some apparatus which can be closed at 

 the required depth. There are many forms which are 

 suitable, and as they are generally made to carry a 

 thermometer the latter may be discussed here. 



There is one rule which is of the greatest importance 

 — that is, that no thermometer is to be used unless 

 its errors are known. All the forms of thermometer 

 used for marine investigation can be bought with a 

 certificate of examination at some recognized in- 

 stitution. In England the examination is made at 

 Kew, in Germany at the Physikalisch-Technische 

 Reichsanstalt at Charlottenburg. It should be re- 

 membered that the error is stated in different ways. 

 The Kew certificate gives the correction to be applied 

 to a given reading in order to find the true temperature. 

 Thus, if the thermometer showed io° and the correc- 

 tion is given as -o-i°, the true reading would be the 

 sum of the observed reading and the correction — that 

 is, 9-9°. The German certificates give the error ; — o-i° 

 here would mean that the thermometer read o-i° 

 too low, so that the true reading would be 10 -i°. The 

 certificate can, of course, only show the errors at the 

 time of examination, and these errors may be called 

 errors of graduation. 



There is, however, another error which is to be 

 guarded against. The glass of the thermometer con- 

 tinues to contract slightly for years after it is made, 

 and so causes a slow rise in the zero. In order to 

 detect this the thermometer should be tested at in- 

 tervals, either by comparison with a standard instru- 

 ment of known accuracy, or better by means of 



