260 



THE ATLANTIC. 



[chap. v. 



son why underlying water miglit not in some cases liave a tem- 

 perature higher than that of the layers above it ; but the ther- 

 mometer is not constructed to show such an anomaly : having 

 once registered its minimum, it has no power of amendment. 



I have no hesitation, therefore, in saying that any single in- 

 dication with a thermometer on Six's principle is not trust- 

 worthy, and that a fact in temperature distribution can only 

 be established by a series of corroborative determinations. 



Although the gross errors to which an unprotected thermom- 

 eter is liable from pressure may be said to be got rid of by the 

 addition of the outer shell, a certain amount of error in the 

 same direction still remains, probably from a slight compres- 

 sion of the unprotected parts of the tube. This error, which 

 is one of slight excess, although for practical purposes it might 

 perhaps be safely regarded as the same for all thermometers, is 

 in detail special to each instrument, and all our thermometers 

 were tested by Captain Tizard, and their individual errors tab- 

 ulated for every 100 fathoms. 



The following table, which is given as an example, is in Fah- 

 renheit degrees : 



These particular thermometers were part of a batch sent out 

 to us late in the cruise, specially strengthened, and certainly 

 of a better construction than those which we had had before. 

 By testing a large series of the earlier instruments in a Bra- 

 mah's press. Captain Davis had come to the conclusion that, 

 when subjected to a pressure corresponding to a depth of 2000 



